Skip to contentSkip to conversation

KVM: s390: enable zPCI for interpretive execution

PATCH Superseded Matthew Rosato 5 participants 97 messages Jan 14, 2022 – Jan 27, 2022

Subsystems (11)
KVM s390 IOMMU VFIO driver +7 more subsystems

v2 of 9 Superseded Latest is v9 →

lore
Share & tools
lore

Matthew Rosato

author maintainer
#0 lore raw

v2 cover KVM: s390: enable zPCI for interpretive execution

Enable interpretive execution of zPCI instructions + adapter interruption forwarding for s390x KVM vfio-pci. This is done by introducing a series of new vfio-pci feature ioctls that are unique vfio-pci-zdev (s390x) and are used to negotiate the various aspects of zPCI interpretation setup. By allowing intepretation of zPCI instructions and firmware delivery of interrupts to guests, we can significantly reduce the frequency of guest SIE exits for zPCI. We then see additional gains by handling a hot-path instruction that can still intercept to the hypervisor (RPCIT) directly in kvm.

From the perspective of guest configuration, you passthrough zPCI devices in the same manner as before, with intepretation support being used by default if available in kernel+qemu.

Will reply with a link to the associated QEMU series.

Changes v1->v2:
- s/has_zpci_interp/has_zpci_lsi/ (Christian)
- Added many R-bs / ACKs (Thanks!)
- Re-work zpci_set_irq_ctrl (Niklas)
- Simplify changes made for zpci_get_mdd (Niklas, Christian)
- 'KVM: s390: pci: add basic kvm_zdev structure' changes (Pierre)
- only build s390/kvm/pci.o when CONFIG_PCI
- Related to the above, add some more checks for IS_ENABLED(CONFIG_PCI) (Pierre)
- Drop set_kvm_facility until VSIE support (Christian)
- Use sclp check instead of stfle when setting ECB (Christian)
- remove unnecessary externs from header (Pierre)
- macro for checkling if shadow ioat is initialized (Pierre)
- Fix interrupt case where we have both AEN and alert list (Christian)
- Re-work AEN setup to satisfy firmware requirements on all supported platforms
- V!=R changes (Niklas)
- vifo_pci_zdev_feat_* - check argz against data size (Pierre, Alex)
- vfio_pci_zdev_{open,release} switch to return void (Alex)
- Related to the above, make kvm_s390_pci_*_probe functions return error if KVM is not registered for the device (Alex)
- Fix my probe implementation to ignore GET|SET, as these don't change the result of the probe. I was erroneously performing the GET or SET operation if specified along with PROBE.
- A few additional fixes regarding ioctl implementation. Return EINVAL if none of PROBE|GET|SET are specified. Return EINVAL if both GET and SET are specified (without PROBE).
- New patch to return status from zpci_refresh_trans (Pierre, Niklas)
- And use that status when possible for KVM rpcit intercept (Pierre)

Matthew Rosato (30):
s390/sclp: detect the zPCI load/store interpretation facility s390/sclp: detect the AISII facility
s390/sclp: detect the AENI facility
s390/sclp: detect the AISI facility
s390/airq: pass more TPI info to airq handlers
s390/airq: allow for airq structure that uses an input vector s390/pci: externalize the SIC operation controls and routine s390/pci: stash associated GISA designation
s390/pci: export some routines related to RPCIT processing s390/pci: stash dtsm and maxstbl
s390/pci: add helper function to find device by handle s390/pci: get SHM information from list pci
s390/pci: return status from zpci_refresh_trans
KVM: s390: pci: add basic kvm_zdev structure
KVM: s390: pci: do initial setup for AEN interpretation
KVM: s390: pci: enable host forwarding of Adapter Event Notifications
KVM: s390: mechanism to enable guest zPCI Interpretation
KVM: s390: pci: provide routines for enabling/disabling interpretation
KVM: s390: pci: provide routines for enabling/disabling interrupt
forwarding
KVM: s390: pci: provide routines for enabling/disabling IOAT assist
KVM: s390: pci: handle refresh of PCI translations
KVM: s390: intercept the rpcit instruction
vfio/pci: re-introduce CONFIG_VFIO_PCI_ZDEV
vfio-pci/zdev: wire up group notifier
vfio-pci/zdev: wire up zPCI interpretive execution support vfio-pci/zdev: wire up zPCI adapter interrupt forwarding support vfio-pci/zdev: wire up zPCI IOAT assist support vfio-pci/zdev: add DTSM to clp group capability
KVM: s390: introduce CPU feature for zPCI Interpretation
MAINTAINERS: additional files related kvm s390 pci passthrough

 MAINTAINERS                      |   2 +
 arch/s390/include/asm/airq.h     |   7 +-
 arch/s390/include/asm/kvm_host.h |   5 +
 arch/s390/include/asm/kvm_pci.h  |  62 +++
 arch/s390/include/asm/pci.h      |  12 +
 arch/s390/include/asm/pci_clp.h  |  11 +-
 arch/s390/include/asm/pci_dma.h  |   3 +
 arch/s390/include/asm/pci_insn.h |  31 +-
 arch/s390/include/asm/sclp.h     |   4 +
 arch/s390/include/asm/tpi.h      |  13 +
 arch/s390/include/uapi/asm/kvm.h |   1 +
 arch/s390/kvm/Makefile           |   2 +-
 arch/s390/kvm/interrupt.c        |  94 +++-
 arch/s390/kvm/kvm-s390.c         |  56 ++-
 arch/s390/kvm/kvm-s390.h         |  10 +
 arch/s390/kvm/pci.c              | 837 +++++++++++++++++++++++++++++++
 arch/s390/kvm/pci.h              |  59 +++
 arch/s390/kvm/priv.c             |  46 ++
 arch/s390/pci/pci.c              |  31 ++
 arch/s390/pci/pci_clp.c          |  31 +-
 arch/s390/pci/pci_dma.c          |   7 +-
 arch/s390/pci/pci_insn.c         |  15 +-
 arch/s390/pci/pci_irq.c          |  48 +-
 drivers/iommu/s390-iommu.c       |   4 +-
 drivers/s390/char/sclp_early.c   |   4 +
 drivers/s390/cio/airq.c          |  12 +-
 drivers/s390/cio/qdio_thinint.c  |   6 +-
 drivers/s390/crypto/ap_bus.c     |   9 +-
 drivers/s390/virtio/virtio_ccw.c |   6 +-
 drivers/vfio/pci/Kconfig         |  11 +
 drivers/vfio/pci/Makefile        |   2 +-
 drivers/vfio/pci/vfio_pci_core.c |   8 +
 drivers/vfio/pci/vfio_pci_zdev.c | 290 ++++++++++-
 include/linux/vfio_pci_core.h    |  42 +-
 include/uapi/linux/vfio.h        |  22 +
 include/uapi/linux/vfio_zdev.h   |  51 ++
 36 files changed, 1788 insertions(+), 66 deletions(-)
 create mode 100644 arch/s390/include/asm/kvm_pci.h
 create mode 100644 arch/s390/kvm/pci.c
 create mode 100644 arch/s390/kvm/pci.h
Signature · 1 line
2.27.0
author maintainer
#1 lore raw

PATCH v2 1/30 s390/sclp: detect the zPCI load/store interpretation facility

Detect the zPCI Load/Store Interpretation facility.

---
 arch/s390/include/asm/sclp.h   | 1 +
 drivers/s390/char/sclp_early.c | 1 +
 2 files changed, 2 insertions(+)
arch/s390/include/asm/sclp.h +1 -0
index c68ea35de498..58a4d3d354b7 100644
@@ -88,6 +88,7 @@ struct sclp_info {
 	unsigned char has_diag318 : 1;
 	unsigned char has_sipl : 1;
 	unsigned char has_dirq : 1;
+	unsigned char has_zpci_lsi : 1;
 	unsigned int ibc;
 	unsigned int mtid;
 	unsigned int mtid_cp;
drivers/s390/char/sclp_early.c +1 -0
index e9943a86c361..b88dd0da1231 100644
@@ -45,6 +45,7 @@ static void __init sclp_early_facilities_detect(void)
 	sclp.has_gisaf = !!(sccb->fac118 & 0x08);
 	sclp.has_hvs = !!(sccb->fac119 & 0x80);
 	sclp.has_kss = !!(sccb->fac98 & 0x01);
+	sclp.has_zpci_lsi = !!(sccb->fac118 & 0x01);
 	if (sccb->fac85 & 0x02)
 		S390_lowcore.machine_flags |= MACHINE_FLAG_ESOP;
 	if (sccb->fac91 & 0x40)
Signature · 1 line
2.27.0
author maintainer
#2 lore raw

PATCH v2 2/30 s390/sclp: detect the AISII facility

In reply to Matthew Rosato

Detect the Adapter Interruption Source ID Interpretation facility.

---
 arch/s390/include/asm/sclp.h   | 1 +
 drivers/s390/char/sclp_early.c | 1 +
 2 files changed, 2 insertions(+)
arch/s390/include/asm/sclp.h +1 -0
index 58a4d3d354b7..8b56ac5ae496 100644
@@ -89,6 +89,7 @@ struct sclp_info {
 	unsigned char has_sipl : 1;
 	unsigned char has_dirq : 1;
 	unsigned char has_zpci_lsi : 1;
+	unsigned char has_aisii : 1;
 	unsigned int ibc;
 	unsigned int mtid;
 	unsigned int mtid_cp;
drivers/s390/char/sclp_early.c +1 -0
index b88dd0da1231..29fee179e197 100644
@@ -45,6 +45,7 @@ static void __init sclp_early_facilities_detect(void)
 	sclp.has_gisaf = !!(sccb->fac118 & 0x08);
 	sclp.has_hvs = !!(sccb->fac119 & 0x80);
 	sclp.has_kss = !!(sccb->fac98 & 0x01);
+	sclp.has_aisii = !!(sccb->fac118 & 0x40);
 	sclp.has_zpci_lsi = !!(sccb->fac118 & 0x01);
 	if (sccb->fac85 & 0x02)
 		S390_lowcore.machine_flags |= MACHINE_FLAG_ESOP;
Signature · 1 line
2.27.0
author maintainer
#3 lore raw

PATCH v2 3/30 s390/sclp: detect the AENI facility

In reply to Matthew Rosato

Detect the Adapter Event Notification Interpretation facility.

---
 arch/s390/include/asm/sclp.h   | 1 +
 drivers/s390/char/sclp_early.c | 1 +
 2 files changed, 2 insertions(+)
arch/s390/include/asm/sclp.h +1 -0
index 8b56ac5ae496..8c2e142000d4 100644
@@ -90,6 +90,7 @@ struct sclp_info {
 	unsigned char has_dirq : 1;
 	unsigned char has_zpci_lsi : 1;
 	unsigned char has_aisii : 1;
+	unsigned char has_aeni : 1;
 	unsigned int ibc;
 	unsigned int mtid;
 	unsigned int mtid_cp;
drivers/s390/char/sclp_early.c +1 -0
index 29fee179e197..e9af01b4c97a 100644
@@ -46,6 +46,7 @@ static void __init sclp_early_facilities_detect(void)
 	sclp.has_hvs = !!(sccb->fac119 & 0x80);
 	sclp.has_kss = !!(sccb->fac98 & 0x01);
 	sclp.has_aisii = !!(sccb->fac118 & 0x40);
+	sclp.has_aeni = !!(sccb->fac118 & 0x20);
 	sclp.has_zpci_lsi = !!(sccb->fac118 & 0x01);
 	if (sccb->fac85 & 0x02)
 		S390_lowcore.machine_flags |= MACHINE_FLAG_ESOP;
Signature · 1 line
2.27.0
author maintainer
#4 lore raw

PATCH v2 4/30 s390/sclp: detect the AISI facility

In reply to Matthew Rosato

Detect the Adapter Interruption Suppression Interpretation facility.

---
 arch/s390/include/asm/sclp.h   | 1 +
 drivers/s390/char/sclp_early.c | 1 +
 2 files changed, 2 insertions(+)
arch/s390/include/asm/sclp.h +1 -0
index 8c2e142000d4..33b174007848 100644
@@ -91,6 +91,7 @@ struct sclp_info {
 	unsigned char has_zpci_lsi : 1;
 	unsigned char has_aisii : 1;
 	unsigned char has_aeni : 1;
+	unsigned char has_aisi : 1;
 	unsigned int ibc;
 	unsigned int mtid;
 	unsigned int mtid_cp;
drivers/s390/char/sclp_early.c +1 -0
index e9af01b4c97a..c13e55cc4a5d 100644
@@ -47,6 +47,7 @@ static void __init sclp_early_facilities_detect(void)
 	sclp.has_kss = !!(sccb->fac98 & 0x01);
 	sclp.has_aisii = !!(sccb->fac118 & 0x40);
 	sclp.has_aeni = !!(sccb->fac118 & 0x20);
+	sclp.has_aisi = !!(sccb->fac118 & 0x10);
 	sclp.has_zpci_lsi = !!(sccb->fac118 & 0x01);
 	if (sccb->fac85 & 0x02)
 		S390_lowcore.machine_flags |= MACHINE_FLAG_ESOP;
Signature · 1 line
2.27.0
#5 lore raw

Re: [PATCH v2 04/30] s390/sclp: detect the AISI facility

Quoted 28 lines · Matthew Rosato

Detect the Adapter Interruption Suppression Interpretation facility.

---
  arch/s390/include/asm/sclp.h   | 1 +
  drivers/s390/char/sclp_early.c | 1 +
  2 files changed, 2 insertions(+)
arch/s390/include/asm/sclp.h +1 -0
index 8c2e142000d4..33b174007848 100644
@@ -91,6 +91,7 @@ struct sclp_info {
  	unsigned char has_zpci_lsi : 1;
  	unsigned char has_aisii : 1;
  	unsigned char has_aeni : 1;
+	unsigned char has_aisi : 1;
  	unsigned int ibc;
  	unsigned int mtid;
  	unsigned int mtid_cp;
drivers/s390/char/sclp_early.c +0 -0
index e9af01b4c97a..c13e55cc4a5d 100644

drivers/s390/char/sclp_early.c static void __init sclp_early_facilities_detect(void) · line 53

Quoted 7 lines
drivers/s390/char/sclp_early.c +1 -0
@@ -47,6 +47,7 @@ static void __init sclp_early_facilities_detect(void)
  	sclp.has_kss = !!(sccb->fac98 & 0x01);
  	sclp.has_aisii = !!(sccb->fac118 & 0x40);
  	sclp.has_aeni = !!(sccb->fac118 & 0x20);
+	sclp.has_aisi = !!(sccb->fac118 & 0x10);
  	sclp.has_zpci_lsi = !!(sccb->fac118 & 0x01);
  	if (sccb->fac85 & 0x02)
  		S390_lowcore.machine_flags |= MACHINE_FLAG_ESOP;

Just a matter of taste, but I'd maybe squash patches 1 - 4 into one patch.

Thomas

author maintainer
#6 lore raw

PATCH v2 5/30 s390/airq: pass more TPI info to airq handlers

In reply to Matthew Rosato

A subsequent patch will introduce an airq handler that requires additional TPI information beyond directed vs floating, so pass the entire tpi_info structure via the handler. Only pci actually uses this information today, for the other airq handlers this is effectively a no-op.

---
 arch/s390/include/asm/airq.h     | 3 ++-
 arch/s390/kvm/interrupt.c        | 4 +++-
 arch/s390/pci/pci_irq.c          | 9 +++++++--
 drivers/s390/cio/airq.c          | 2 +-
 drivers/s390/cio/qdio_thinint.c  | 6 ++++--
 drivers/s390/crypto/ap_bus.c     | 9 ++++++---
 drivers/s390/virtio/virtio_ccw.c | 4 +++-
 7 files changed, 26 insertions(+), 11 deletions(-)
arch/s390/include/asm/airq.h +2 -1
index 01936fdfaddb..7918a7d09028 100644
@@ -12,10 +12,11 @@
 
 #include <linux/bit_spinlock.h>
 #include <linux/dma-mapping.h>
+#include <asm/tpi.h>
 
 struct airq_struct {
 	struct hlist_node list;		/* Handler queueing. */
-	void (*handler)(struct airq_struct *airq, bool floating);
+	void (*handler)(struct airq_struct *airq, struct tpi_info *tpi_info);
 	u8 *lsi_ptr;			/* Local-Summary-Indicator pointer */
 	u8 lsi_mask;			/* Local-Summary-Indicator mask */
 	u8 isc;				/* Interrupt-subclass */
arch/s390/kvm/interrupt.c +3 -1
index c3bd993fdd0c..f9b872e358c6 100644
@@ -28,6 +28,7 @@
 #include <asm/switch_to.h>
 #include <asm/nmi.h>
 #include <asm/airq.h>
+#include <asm/tpi.h>
 #include "kvm-s390.h"
 #include "gaccess.h"
 #include "trace-s390.h"
@@ -3261,7 +3262,8 @@ int kvm_s390_gisc_unregister(struct kvm *kvm, u32 gisc)
 }
 EXPORT_SYMBOL_GPL(kvm_s390_gisc_unregister);
 
-static void gib_alert_irq_handler(struct airq_struct *airq, bool floating)
+static void gib_alert_irq_handler(struct airq_struct *airq,
+				  struct tpi_info *tpi_info)
 {
 	inc_irq_stat(IRQIO_GAL);
 	process_gib_alert_list();
arch/s390/pci/pci_irq.c +7 -2
index 2b6062c486f5..cc4c8d7c8f5c 100644
@@ -11,6 +11,7 @@
 
 #include <asm/isc.h>
 #include <asm/airq.h>
+#include <asm/tpi.h>
 
 static enum {FLOATING, DIRECTED} irq_delivery;
 
@@ -216,8 +217,11 @@ static void zpci_handle_fallback_irq(void)
 	}
 }
 
-static void zpci_directed_irq_handler(struct airq_struct *airq, bool floating)
+static void zpci_directed_irq_handler(struct airq_struct *airq,
+				      struct tpi_info *tpi_info)
 {
+	bool floating = !tpi_info->directed_irq;
+
 	if (floating) {
 		inc_irq_stat(IRQIO_PCF);
 		zpci_handle_fallback_irq();
@@ -227,7 +231,8 @@ static void zpci_directed_irq_handler(struct airq_struct *airq, bool floating)
 	}
 }
 
-static void zpci_floating_irq_handler(struct airq_struct *airq, bool floating)
+static void zpci_floating_irq_handler(struct airq_struct *airq,
+				      struct tpi_info *tpi_info)
 {
 	unsigned long si, ai;
 	struct airq_iv *aibv;
drivers/s390/cio/airq.c +1 -1
index e56535c99888..2f2226786319 100644
@@ -99,7 +99,7 @@ static irqreturn_t do_airq_interrupt(int irq, void *dummy)
 	rcu_read_lock();
 	hlist_for_each_entry_rcu(airq, head, list)
 		if ((*airq->lsi_ptr & airq->lsi_mask) != 0)
-			airq->handler(airq, !tpi_info->directed_irq);
+			airq->handler(airq, tpi_info);
 	rcu_read_unlock();
 
 	return IRQ_HANDLED;
drivers/s390/cio/qdio_thinint.c +4 -2
index 8e09bf3a2fcd..9b9335dd06db 100644
@@ -15,6 +15,7 @@
 #include <asm/qdio.h>
 #include <asm/airq.h>
 #include <asm/isc.h>
+#include <asm/tpi.h>
 
 #include "cio.h"
 #include "ioasm.h"
@@ -93,9 +94,10 @@ static inline u32 clear_shared_ind(void)
 /**
  * tiqdio_thinint_handler - thin interrupt handler for qdio
  * @airq: pointer to adapter interrupt descriptor
- * @floating: flag to recognize floating vs. directed interrupts (unused)
+ * @tpi_info: interrupt information (e.g. floating vs directed -- unused)
  */
-static void tiqdio_thinint_handler(struct airq_struct *airq, bool floating)
+static void tiqdio_thinint_handler(struct airq_struct *airq,
+				   struct tpi_info *tpi_info)
 {
 	u64 irq_time = S390_lowcore.int_clock;
 	u32 si_used = clear_shared_ind();
drivers/s390/crypto/ap_bus.c +6 -3
index 1986243f9cd3..df1a038442db 100644
@@ -27,6 +27,7 @@
 #include <linux/kthread.h>
 #include <linux/mutex.h>
 #include <asm/airq.h>
+#include <asm/tpi.h>
 #include <linux/atomic.h>
 #include <asm/isc.h>
 #include <linux/hrtimer.h>
@@ -129,7 +130,8 @@ static int ap_max_adapter_id = 63;
 static struct bus_type ap_bus_type;
 
 /* Adapter interrupt definitions */
-static void ap_interrupt_handler(struct airq_struct *airq, bool floating);
+static void ap_interrupt_handler(struct airq_struct *airq,
+				 struct tpi_info *tpi_info);
 
 static bool ap_irq_flag;
 
@@ -442,9 +444,10 @@ static enum hrtimer_restart ap_poll_timeout(struct hrtimer *unused)
 /**
  * ap_interrupt_handler() - Schedule ap_tasklet on interrupt
  * @airq: pointer to adapter interrupt descriptor
- * @floating: ignored
+ * @tpi_info: ignored
  */
-static void ap_interrupt_handler(struct airq_struct *airq, bool floating)
+static void ap_interrupt_handler(struct airq_struct *airq,
+				 struct tpi_info *tpi_info)
 {
 	inc_irq_stat(IRQIO_APB);
 	tasklet_schedule(&ap_tasklet);
drivers/s390/virtio/virtio_ccw.c +3 -1
index d35e7a3f7067..52c376d15978 100644
@@ -33,6 +33,7 @@
 #include <asm/virtio-ccw.h>
 #include <asm/isc.h>
 #include <asm/airq.h>
+#include <asm/tpi.h>
 
 /*
  * virtio related functions
@@ -203,7 +204,8 @@ static void drop_airq_indicator(struct virtqueue *vq, struct airq_info *info)
 	write_unlock_irqrestore(&info->lock, flags);
 }
 
-static void virtio_airq_handler(struct airq_struct *airq, bool floating)
+static void virtio_airq_handler(struct airq_struct *airq,
+				struct tpi_info *tpi_info)
 {
 	struct airq_info *info = container_of(airq, struct airq_info, airq);
 	unsigned long ai;
Signature · 1 line
2.27.0
Reviewed-by on v2 5/30
#7 lore raw

Re: [PATCH v2 05/30] s390/airq: pass more TPI info to airq handlers

Quoted 10 lines · Matthew Rosato

A subsequent patch will introduce an airq handler that requires additional TPI information beyond directed vs floating, so pass the entire tpi_info structure via the handler. Only pci actually uses this information today, for the other airq handlers this is effectively a no-op.

Quoted 9 lines
---
  arch/s390/include/asm/airq.h     | 3 ++-
  arch/s390/kvm/interrupt.c        | 4 +++-
  arch/s390/pci/pci_irq.c          | 9 +++++++--
  drivers/s390/cio/airq.c          | 2 +-
  drivers/s390/cio/qdio_thinint.c  | 6 ++++--
  drivers/s390/crypto/ap_bus.c     | 9 ++++++---
  drivers/s390/virtio/virtio_ccw.c | 4 +++-
  7 files changed, 26 insertions(+), 11 deletions(-)
author maintainer
#8 lore raw

PATCH v2 6/30 s390/airq: allow for airq structure that uses an input vector

In reply to Matthew Rosato

When doing device passthrough where interrupts are being forwarded from host to guest, we wish to use a pinned section of guest memory as the vector (the same memory used by the guest as the vector).

---
 arch/s390/include/asm/airq.h     |  4 +++-
 arch/s390/pci/pci_irq.c          |  8 ++++----
 drivers/s390/cio/airq.c          | 10 +++++++---
 drivers/s390/virtio/virtio_ccw.c |  2 +-
 4 files changed, 15 insertions(+), 9 deletions(-)
arch/s390/include/asm/airq.h +3 -1
index 7918a7d09028..e82e5626e139 100644
@@ -47,8 +47,10 @@ struct airq_iv {
 #define AIRQ_IV_PTR		4	/* Allocate the ptr array */
 #define AIRQ_IV_DATA		8	/* Allocate the data array */
 #define AIRQ_IV_CACHELINE	16	/* Cacheline alignment for the vector */
+#define AIRQ_IV_GUESTVEC	32	/* Vector is a pinned guest page */
 
-struct airq_iv *airq_iv_create(unsigned long bits, unsigned long flags);
+struct airq_iv *airq_iv_create(unsigned long bits, unsigned long flags,
+			       unsigned long *vec);
 void airq_iv_release(struct airq_iv *iv);
 unsigned long airq_iv_alloc(struct airq_iv *iv, unsigned long num);
 void airq_iv_free(struct airq_iv *iv, unsigned long bit, unsigned long num);
arch/s390/pci/pci_irq.c +4 -4
index cc4c8d7c8f5c..0d0a02a9fbbf 100644
@@ -296,7 +296,7 @@ int arch_setup_msi_irqs(struct pci_dev *pdev, int nvec, int type)
 		zdev->aisb = bit;
 
 		/* Create adapter interrupt vector */
-		zdev->aibv = airq_iv_create(msi_vecs, AIRQ_IV_DATA | AIRQ_IV_BITLOCK);
+		zdev->aibv = airq_iv_create(msi_vecs, AIRQ_IV_DATA | AIRQ_IV_BITLOCK, NULL);
 		if (!zdev->aibv)
 			return -ENOMEM;
 
@@ -419,7 +419,7 @@ static int __init zpci_directed_irq_init(void)
 	union zpci_sic_iib iib = {{0}};
 	unsigned int cpu;
 
-	zpci_sbv = airq_iv_create(num_possible_cpus(), 0);
+	zpci_sbv = airq_iv_create(num_possible_cpus(), 0, NULL);
 	if (!zpci_sbv)
 		return -ENOMEM;
 
@@ -441,7 +441,7 @@ static int __init zpci_directed_irq_init(void)
 		zpci_ibv[cpu] = airq_iv_create(cache_line_size() * BITS_PER_BYTE,
 					       AIRQ_IV_DATA |
 					       AIRQ_IV_CACHELINE |
-					       (!cpu ? AIRQ_IV_ALLOC : 0));
+					       (!cpu ? AIRQ_IV_ALLOC : 0), NULL);
 		if (!zpci_ibv[cpu])
 			return -ENOMEM;
 	}
@@ -458,7 +458,7 @@ static int __init zpci_floating_irq_init(void)
 	if (!zpci_ibv)
 		return -ENOMEM;
 
-	zpci_sbv = airq_iv_create(ZPCI_NR_DEVICES, AIRQ_IV_ALLOC);
+	zpci_sbv = airq_iv_create(ZPCI_NR_DEVICES, AIRQ_IV_ALLOC, NULL);
 	if (!zpci_sbv)
 		goto out_free;
 
drivers/s390/cio/airq.c +7 -3
index 2f2226786319..375a58b1c838 100644
@@ -122,10 +122,12 @@ static inline unsigned long iv_size(unsigned long bits)
  * airq_iv_create - create an interrupt vector
  * @bits: number of bits in the interrupt vector
  * @flags: allocation flags
+ * @vec: pointer to pinned guest memory if AIRQ_IV_GUESTVEC
  *
  * Returns a pointer to an interrupt vector structure
  */
-struct airq_iv *airq_iv_create(unsigned long bits, unsigned long flags)
+struct airq_iv *airq_iv_create(unsigned long bits, unsigned long flags,
+			       unsigned long *vec)
 {
 	struct airq_iv *iv;
 	unsigned long size;
@@ -146,6 +148,8 @@ struct airq_iv *airq_iv_create(unsigned long bits, unsigned long flags)
 					     &iv->vector_dma);
 		if (!iv->vector)
 			goto out_free;
+	} else if (flags & AIRQ_IV_GUESTVEC) {
+		iv->vector = vec;
 	} else {
 		iv->vector = cio_dma_zalloc(size);
 		if (!iv->vector)
@@ -185,7 +189,7 @@ struct airq_iv *airq_iv_create(unsigned long bits, unsigned long flags)
 	kfree(iv->avail);
 	if (iv->flags & AIRQ_IV_CACHELINE && iv->vector)
 		dma_pool_free(airq_iv_cache, iv->vector, iv->vector_dma);
-	else
+	else if (!(iv->flags & AIRQ_IV_GUESTVEC))
 		cio_dma_free(iv->vector, size);
 	kfree(iv);
 out:
@@ -204,7 +208,7 @@ void airq_iv_release(struct airq_iv *iv)
 	kfree(iv->bitlock);
 	if (iv->flags & AIRQ_IV_CACHELINE)
 		dma_pool_free(airq_iv_cache, iv->vector, iv->vector_dma);
-	else
+	else if (!(iv->flags & AIRQ_IV_GUESTVEC))
 		cio_dma_free(iv->vector, iv_size(iv->bits));
 	kfree(iv->avail);
 	kfree(iv);
drivers/s390/virtio/virtio_ccw.c +1 -1
index 52c376d15978..410498d693f8 100644
@@ -241,7 +241,7 @@ static struct airq_info *new_airq_info(int index)
 		return NULL;
 	rwlock_init(&info->lock);
 	info->aiv = airq_iv_create(VIRTIO_IV_BITS, AIRQ_IV_ALLOC | AIRQ_IV_PTR
-				   | AIRQ_IV_CACHELINE);
+				   | AIRQ_IV_CACHELINE, NULL);
 	if (!info->aiv) {
 		kfree(info);
 		return NULL;
Signature · 1 line
2.27.0
maintainer
#9 lore raw

Re: [PATCH v2 06/30] s390/airq: allow for airq structure that uses an input vector

Quoted 3 lines · On Fri, 14 Jan 2022 15:31:21 -0500 Matthew Rosato <[email protected]>…

When doing device passthrough where interrupts are being forwarded from host to guest, we wish to use a pinned section of guest memory as the vector (the same memory used by the guest as the vector).

maybe expand the description of the patch to explain what exactly is being done in this patch. Namely: you add a parameter to a function (and some logic in the function to use the new parameter), but the function is not being used yet. And pinning is also done somewhere else.

maybe you can add something like

This patch adds a new parameter for airq_iv_create to pass the existing vector pinned in guest memory and to use it when needed instead of allocating a new one.

Apart from that, the patch looks good.

Quoted 131 lines
---
 arch/s390/include/asm/airq.h     |  4 +++-
 arch/s390/pci/pci_irq.c          |  8 ++++----
 drivers/s390/cio/airq.c          | 10 +++++++---
 drivers/s390/virtio/virtio_ccw.c |  2 +-
 4 files changed, 15 insertions(+), 9 deletions(-)
arch/s390/include/asm/airq.h +3 -1
index 7918a7d09028..e82e5626e139 100644
@@ -47,8 +47,10 @@ struct airq_iv {
 #define AIRQ_IV_PTR		4	/* Allocate the ptr array */
 #define AIRQ_IV_DATA		8	/* Allocate the data array */
 #define AIRQ_IV_CACHELINE	16	/* Cacheline alignment for the vector */
+#define AIRQ_IV_GUESTVEC	32	/* Vector is a pinned guest page */
 
-struct airq_iv *airq_iv_create(unsigned long bits, unsigned long flags);
+struct airq_iv *airq_iv_create(unsigned long bits, unsigned long
 flags,
+			       unsigned long *vec);
 void airq_iv_release(struct airq_iv *iv);
 unsigned long airq_iv_alloc(struct airq_iv *iv, unsigned long num);

void airq_iv_free(struct airq_iv *iv, unsigned long bit, unsigned
long num); diff --git a/arch/s390/pci/pci_irq.c b/arch/s390/pci/pci_irq.c index cc4c8d7c8f5c..0d0a02a9fbbf 100644

arch/s390/pci/pci_irq.c +2 -2
@@ -296,7 +296,7 @@ int arch_setup_msi_irqs(struct pci_dev *pdev, int
 nvec, int type) zdev->aisb = bit;
 
 		/* Create adapter interrupt vector */
-		zdev->aibv = airq_iv_create(msi_vecs, AIRQ_IV_DATA |
 AIRQ_IV_BITLOCK);
+		zdev->aibv = airq_iv_create(msi_vecs, AIRQ_IV_DATA |
 AIRQ_IV_BITLOCK, NULL); if (!zdev->aibv)
 			return -ENOMEM;
@@ -419,7 +419,7 @@ static int __init zpci_directed_irq_init(void)
 	union zpci_sic_iib iib = {{0}};
 	unsigned int cpu;
 
-	zpci_sbv = airq_iv_create(num_possible_cpus(), 0);
+	zpci_sbv = airq_iv_create(num_possible_cpus(), 0, NULL);
 	if (!zpci_sbv)
 		return -ENOMEM;
 
@@ -441,7 +441,7 @@ static int __init zpci_directed_irq_init(void)
 		zpci_ibv[cpu] = airq_iv_create(cache_line_size() *

BITS_PER_BYTE, AIRQ_IV_DATA |

arch/s390/pci/pci_irq.c +0 -1
 					       AIRQ_IV_CACHELINE |
-					       (!cpu ? AIRQ_IV_ALLOC

: 0));

arch/s390/pci/pci_irq.c +1 -0
+					       (!cpu ? AIRQ_IV_ALLOC

: 0), NULL); if (!zpci_ibv[cpu])

arch/s390/pci/pci_irq.c +1 -1
 			return -ENOMEM;
 	}
@@ -458,7 +458,7 @@ static int __init zpci_floating_irq_init(void)
 	if (!zpci_ibv)
 		return -ENOMEM;
 
-	zpci_sbv = airq_iv_create(ZPCI_NR_DEVICES, AIRQ_IV_ALLOC);
+	zpci_sbv = airq_iv_create(ZPCI_NR_DEVICES, AIRQ_IV_ALLOC,
 NULL); if (!zpci_sbv)
 		goto out_free;
 
drivers/s390/cio/airq.c +1 -1
index 2f2226786319..375a58b1c838 100644
@@ -122,10 +122,12 @@ static inline unsigned long iv_size(unsigned
 long bits)
  * airq_iv_create - create an interrupt vector
  * @bits: number of bits in the interrupt vector
  * @flags: allocation flags
+ * @vec: pointer to pinned guest memory if AIRQ_IV_GUESTVEC
  *
  * Returns a pointer to an interrupt vector structure
  */
-struct airq_iv *airq_iv_create(unsigned long bits, unsigned long

flags) +struct airq_iv *airq_iv_create(unsigned long bits, unsigned long flags,

drivers/s390/cio/airq.c +4 -1
+			       unsigned long *vec)
 {
 	struct airq_iv *iv;
 	unsigned long size;
@@ -146,6 +148,8 @@ struct airq_iv *airq_iv_create(unsigned long
 bits, unsigned long flags) &iv->vector_dma);
 		if (!iv->vector)
 			goto out_free;
+	} else if (flags & AIRQ_IV_GUESTVEC) {
+		iv->vector = vec;
 	} else {
 		iv->vector = cio_dma_zalloc(size);
 		if (!iv->vector)
@@ -185,7 +189,7 @@ struct airq_iv *airq_iv_create(unsigned long
 bits, unsigned long flags) kfree(iv->avail);
 	if (iv->flags & AIRQ_IV_CACHELINE && iv->vector)
 		dma_pool_free(airq_iv_cache, iv->vector,
 iv->vector_dma);
-	else
+	else if (!(iv->flags & AIRQ_IV_GUESTVEC))
 		cio_dma_free(iv->vector, size);
 	kfree(iv);

out:

drivers/s390/cio/airq.c +1 -1
@@ -204,7 +208,7 @@ void airq_iv_release(struct airq_iv *iv)
 	kfree(iv->bitlock);
 	if (iv->flags & AIRQ_IV_CACHELINE)
 		dma_pool_free(airq_iv_cache, iv->vector,
 iv->vector_dma);
-	else
+	else if (!(iv->flags & AIRQ_IV_GUESTVEC))
 		cio_dma_free(iv->vector, iv_size(iv->bits));
 	kfree(iv->avail);
drivers/s390/cio/airq.c +0 -0
 	kfree(iv);
(quoted diff) +0 -0

b/drivers/s390/virtio/virtio_ccw.c index 52c376d15978..410498d693f8 100644 --- a/drivers/s390/virtio/virtio_ccw.c
+++ b/drivers/s390/virtio/virtio_ccw.c

drivers/s390/virtio/virtio_ccw.c +0 -0
@@ -241,7 +241,7 @@ static struct airq_info *new_airq_info(int index)
 		return NULL;
 	rwlock_init(&info->lock);
 	info->aiv = airq_iv_create(VIRTIO_IV_BITS, AIRQ_IV_ALLOC |

AIRQ_IV_PTR

drivers/s390/virtio/virtio_ccw.c +1 -1
-				   | AIRQ_IV_CACHELINE);
+				   | AIRQ_IV_CACHELINE, NULL);
 	if (!info->aiv) {
 		kfree(info);
 		return NULL;
author maintainer
#10 lore raw

Re: [PATCH v2 06/30] s390/airq: allow for airq structure that uses an input vector

Quoted 12 lines · Claudio Imbrenda
Quoted 3 lines · On Fri, 14 Jan 2022 15:31:21 -0500 Matthew Rosato <[email protected]>…

When doing device passthrough where interrupts are being forwarded from host to guest, we wish to use a pinned section of guest memory as the vector (the same memory used by the guest as the vector).

maybe expand the description of the patch to explain what exactly is being done in this patch. Namely: you add a parameter to a function (and some logic in the function to use the new parameter), but the function is not being used yet. And pinning is also done somewhere else.

Quoted 8 lines

maybe you can add something like

This patch adds a new parameter for airq_iv_create to pass the existing vector pinned in guest memory and to use it when needed instead of allocating a new one.

Apart from that, the patch looks good.

Thanks, will re-work to:

When doing device passthrough where interrupts are being forwarded from host to guest, we wish to use a pinned section of guest memory as the vector (the same memory used by the guest as the vector).
To accomplish this, add a new parameter for airq_iv_create which allows passing an existing vector to be used instead of allocating a new one. The caller is responsible for ensuring the vector is pinned in memory as well as for unpinning the memory when the vector is no longer needed. A subsequent patch will use this new parameter for zPCI interpretation.

Reviewed-by on v2 6/30
#11 lore raw

Re: [PATCH v2 06/30] s390/airq: allow for airq structure that uses an input vector

In reply to Matthew Rosato
Quoted 5 lines · Matthew Rosato

When doing device passthrough where interrupts are being forwarded from host to guest, we wish to use a pinned section of guest memory as the vector (the same memory used by the guest as the vector).

Quoted 122 lines
---
  arch/s390/include/asm/airq.h     |  4 +++-
  arch/s390/pci/pci_irq.c          |  8 ++++----
  drivers/s390/cio/airq.c          | 10 +++++++---
  drivers/s390/virtio/virtio_ccw.c |  2 +-
  4 files changed, 15 insertions(+), 9 deletions(-)
arch/s390/include/asm/airq.h +3 -1
index 7918a7d09028..e82e5626e139 100644
@@ -47,8 +47,10 @@ struct airq_iv {
  #define AIRQ_IV_PTR		4	/* Allocate the ptr array */
  #define AIRQ_IV_DATA		8	/* Allocate the data array */
  #define AIRQ_IV_CACHELINE	16	/* Cacheline alignment for the vector */
+#define AIRQ_IV_GUESTVEC	32	/* Vector is a pinned guest page */
 
-struct airq_iv *airq_iv_create(unsigned long bits, unsigned long flags);
+struct airq_iv *airq_iv_create(unsigned long bits, unsigned long flags,
+			       unsigned long *vec);
  void airq_iv_release(struct airq_iv *iv);
  unsigned long airq_iv_alloc(struct airq_iv *iv, unsigned long num);
  void airq_iv_free(struct airq_iv *iv, unsigned long bit, unsigned long num);
arch/s390/pci/pci_irq.c +4 -4
index cc4c8d7c8f5c..0d0a02a9fbbf 100644
@@ -296,7 +296,7 @@ int arch_setup_msi_irqs(struct pci_dev *pdev, int nvec, int type)
  		zdev->aisb = bit;
 
  		/* Create adapter interrupt vector */
-		zdev->aibv = airq_iv_create(msi_vecs, AIRQ_IV_DATA | AIRQ_IV_BITLOCK);
+		zdev->aibv = airq_iv_create(msi_vecs, AIRQ_IV_DATA | AIRQ_IV_BITLOCK, NULL);
  		if (!zdev->aibv)
  			return -ENOMEM;
 
@@ -419,7 +419,7 @@ static int __init zpci_directed_irq_init(void)
  	union zpci_sic_iib iib = {{0}};
  	unsigned int cpu;
 
-	zpci_sbv = airq_iv_create(num_possible_cpus(), 0);
+	zpci_sbv = airq_iv_create(num_possible_cpus(), 0, NULL);
  	if (!zpci_sbv)
  		return -ENOMEM;
 
@@ -441,7 +441,7 @@ static int __init zpci_directed_irq_init(void)
  		zpci_ibv[cpu] = airq_iv_create(cache_line_size() * BITS_PER_BYTE,
  					       AIRQ_IV_DATA |
  					       AIRQ_IV_CACHELINE |
-					       (!cpu ? AIRQ_IV_ALLOC : 0));
+					       (!cpu ? AIRQ_IV_ALLOC : 0), NULL);
  		if (!zpci_ibv[cpu])
  			return -ENOMEM;
  	}
@@ -458,7 +458,7 @@ static int __init zpci_floating_irq_init(void)
  	if (!zpci_ibv)
  		return -ENOMEM;
 
-	zpci_sbv = airq_iv_create(ZPCI_NR_DEVICES, AIRQ_IV_ALLOC);
+	zpci_sbv = airq_iv_create(ZPCI_NR_DEVICES, AIRQ_IV_ALLOC, NULL);
  	if (!zpci_sbv)
  		goto out_free;
 
drivers/s390/cio/airq.c +7 -3
index 2f2226786319..375a58b1c838 100644
@@ -122,10 +122,12 @@ static inline unsigned long iv_size(unsigned long bits)
   * airq_iv_create - create an interrupt vector
   * @bits: number of bits in the interrupt vector
   * @flags: allocation flags
+ * @vec: pointer to pinned guest memory if AIRQ_IV_GUESTVEC
   *
   * Returns a pointer to an interrupt vector structure
   */
-struct airq_iv *airq_iv_create(unsigned long bits, unsigned long flags)
+struct airq_iv *airq_iv_create(unsigned long bits, unsigned long flags,
+			       unsigned long *vec)
  {
  	struct airq_iv *iv;
  	unsigned long size;
@@ -146,6 +148,8 @@ struct airq_iv *airq_iv_create(unsigned long bits, unsigned long flags)
  					     &iv->vector_dma);
  		if (!iv->vector)
  			goto out_free;
+	} else if (flags & AIRQ_IV_GUESTVEC) {
+		iv->vector = vec;
  	} else {
  		iv->vector = cio_dma_zalloc(size);
  		if (!iv->vector)
@@ -185,7 +189,7 @@ struct airq_iv *airq_iv_create(unsigned long bits, unsigned long flags)
  	kfree(iv->avail);
  	if (iv->flags & AIRQ_IV_CACHELINE && iv->vector)
  		dma_pool_free(airq_iv_cache, iv->vector, iv->vector_dma);
-	else
+	else if (!(iv->flags & AIRQ_IV_GUESTVEC))
  		cio_dma_free(iv->vector, size);
  	kfree(iv);
  out:
@@ -204,7 +208,7 @@ void airq_iv_release(struct airq_iv *iv)
  	kfree(iv->bitlock);
  	if (iv->flags & AIRQ_IV_CACHELINE)
  		dma_pool_free(airq_iv_cache, iv->vector, iv->vector_dma);
-	else
+	else if (!(iv->flags & AIRQ_IV_GUESTVEC))
  		cio_dma_free(iv->vector, iv_size(iv->bits));
  	kfree(iv->avail);
  	kfree(iv);
drivers/s390/virtio/virtio_ccw.c +1 -1
index 52c376d15978..410498d693f8 100644
@@ -241,7 +241,7 @@ static struct airq_info *new_airq_info(int index)
  		return NULL;
  	rwlock_init(&info->lock);
  	info->aiv = airq_iv_create(VIRTIO_IV_BITS, AIRQ_IV_ALLOC | AIRQ_IV_PTR
-				   | AIRQ_IV_CACHELINE);
+				   | AIRQ_IV_CACHELINE, NULL);
  	if (!info->aiv) {
  		kfree(info);
  		return NULL;
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#12 lore raw

PATCH v2 7/30 s390/pci: externalize the SIC operation controls and routine

In reply to Matthew Rosato

A subsequent patch will be issuing SIC from KVM -- export the necessary routine and make the operation control definitions available from a header. Because the routine will now be exported, let's rename __zpci_set_irq_ctrl to zpci_set_irq_ctrl and get rid of the zero'd iib wrapper function of the same name.

---
 arch/s390/include/asm/pci_insn.h | 17 +++++++++--------
 arch/s390/pci/pci_insn.c         |  3 ++-
 arch/s390/pci/pci_irq.c          | 26 ++++++++++++--------------
 3 files changed, 23 insertions(+), 23 deletions(-)
arch/s390/include/asm/pci_insn.h +9 -8
index 61cf9531f68f..5331082fa516 100644
@@ -98,6 +98,14 @@ struct zpci_fib {
 	u32 gd;
 } __packed __aligned(8);
 
+/* Set Interruption Controls Operation Controls  */
+#define	SIC_IRQ_MODE_ALL		0
+#define	SIC_IRQ_MODE_SINGLE		1
+#define	SIC_IRQ_MODE_DIRECT		4
+#define	SIC_IRQ_MODE_D_ALL		16
+#define	SIC_IRQ_MODE_D_SINGLE		17
+#define	SIC_IRQ_MODE_SET_CPU		18
+
 /* directed interruption information block */
 struct zpci_diib {
 	u32 : 1;
@@ -134,13 +142,6 @@ int __zpci_store(u64 data, u64 req, u64 offset);
 int zpci_store(const volatile void __iomem *addr, u64 data, unsigned long len);
 int __zpci_store_block(const u64 *data, u64 req, u64 offset);
 void zpci_barrier(void);
-int __zpci_set_irq_ctrl(u16 ctl, u8 isc, union zpci_sic_iib *iib);
-
-static inline int zpci_set_irq_ctrl(u16 ctl, u8 isc)
-{
-	union zpci_sic_iib iib = {{0}};
-
-	return __zpci_set_irq_ctrl(ctl, isc, &iib);
-}
+int zpci_set_irq_ctrl(u16 ctl, u8 isc, union zpci_sic_iib *iib);
 
 #endif
arch/s390/pci/pci_insn.c +2 -1
index 4dd58b196cea..2a47b3936e44 100644
@@ -97,7 +97,7 @@ int zpci_refresh_trans(u64 fn, u64 addr, u64 range)
 }
 
 /* Set Interruption Controls */
-int __zpci_set_irq_ctrl(u16 ctl, u8 isc, union zpci_sic_iib *iib)
+int zpci_set_irq_ctrl(u16 ctl, u8 isc, union zpci_sic_iib *iib)
 {
 	if (!test_facility(72))
 		return -EIO;
@@ -108,6 +108,7 @@ int __zpci_set_irq_ctrl(u16 ctl, u8 isc, union zpci_sic_iib *iib)
 
 	return 0;
 }
+EXPORT_SYMBOL_GPL(zpci_set_irq_ctrl);
 
 /* PCI Load */
 static inline int ____pcilg(u64 *data, u64 req, u64 offset, u8 *status)
arch/s390/pci/pci_irq.c +12 -14
index 0d0a02a9fbbf..2f675355fd0c 100644
@@ -15,13 +15,6 @@
 
 static enum {FLOATING, DIRECTED} irq_delivery;
 
-#define	SIC_IRQ_MODE_ALL		0
-#define	SIC_IRQ_MODE_SINGLE		1
-#define	SIC_IRQ_MODE_DIRECT		4
-#define	SIC_IRQ_MODE_D_ALL		16
-#define	SIC_IRQ_MODE_D_SINGLE		17
-#define	SIC_IRQ_MODE_SET_CPU		18
-
 /*
  * summary bit vector
  * FLOATING - summary bit per function
@@ -154,6 +147,7 @@ static struct irq_chip zpci_irq_chip = {
 static void zpci_handle_cpu_local_irq(bool rescan)
 {
 	struct airq_iv *dibv = zpci_ibv[smp_processor_id()];
+	union zpci_sic_iib iib = {{0}};
 	unsigned long bit;
 	int irqs_on = 0;
 
@@ -165,7 +159,7 @@ static void zpci_handle_cpu_local_irq(bool rescan)
 				/* End of second scan with interrupts on. */
 				break;
 			/* First scan complete, reenable interrupts. */
-			if (zpci_set_irq_ctrl(SIC_IRQ_MODE_D_SINGLE, PCI_ISC))
+			if (zpci_set_irq_ctrl(SIC_IRQ_MODE_D_SINGLE, PCI_ISC, &iib))
 				break;
 			bit = 0;
 			continue;
@@ -193,6 +187,7 @@ static void zpci_handle_remote_irq(void *data)
 static void zpci_handle_fallback_irq(void)
 {
 	struct cpu_irq_data *cpu_data;
+	union zpci_sic_iib iib = {{0}};
 	unsigned long cpu;
 	int irqs_on = 0;
 
@@ -203,7 +198,7 @@ static void zpci_handle_fallback_irq(void)
 				/* End of second scan with interrupts on. */
 				break;
 			/* First scan complete, reenable interrupts. */
-			if (zpci_set_irq_ctrl(SIC_IRQ_MODE_SINGLE, PCI_ISC))
+			if (zpci_set_irq_ctrl(SIC_IRQ_MODE_SINGLE, PCI_ISC, &iib))
 				break;
 			cpu = 0;
 			continue;
@@ -234,6 +229,7 @@ static void zpci_directed_irq_handler(struct airq_struct *airq,
 static void zpci_floating_irq_handler(struct airq_struct *airq,
 				      struct tpi_info *tpi_info)
 {
+	union zpci_sic_iib iib = {{0}};
 	unsigned long si, ai;
 	struct airq_iv *aibv;
 	int irqs_on = 0;
@@ -247,7 +243,7 @@ static void zpci_floating_irq_handler(struct airq_struct *airq,
 				/* End of second scan with interrupts on. */
 				break;
 			/* First scan complete, reenable interrupts. */
-			if (zpci_set_irq_ctrl(SIC_IRQ_MODE_SINGLE, PCI_ISC))
+			if (zpci_set_irq_ctrl(SIC_IRQ_MODE_SINGLE, PCI_ISC, &iib))
 				break;
 			si = 0;
 			continue;
@@ -407,11 +403,12 @@ static struct airq_struct zpci_airq = {
 static void __init cpu_enable_directed_irq(void *unused)
 {
 	union zpci_sic_iib iib = {{0}};
+	union zpci_sic_iib ziib = {{0}};
 
 	iib.cdiib.dibv_addr = (u64) zpci_ibv[smp_processor_id()]->vector;
 
-	__zpci_set_irq_ctrl(SIC_IRQ_MODE_SET_CPU, 0, &iib);
-	zpci_set_irq_ctrl(SIC_IRQ_MODE_D_SINGLE, PCI_ISC);
+	zpci_set_irq_ctrl(SIC_IRQ_MODE_SET_CPU, 0, &iib);
+	zpci_set_irq_ctrl(SIC_IRQ_MODE_D_SINGLE, PCI_ISC, &ziib);
 }
 
 static int __init zpci_directed_irq_init(void)
@@ -426,7 +423,7 @@ static int __init zpci_directed_irq_init(void)
 	iib.diib.isc = PCI_ISC;
 	iib.diib.nr_cpus = num_possible_cpus();
 	iib.diib.disb_addr = virt_to_phys(zpci_sbv->vector);
-	__zpci_set_irq_ctrl(SIC_IRQ_MODE_DIRECT, 0, &iib);
+	zpci_set_irq_ctrl(SIC_IRQ_MODE_DIRECT, 0, &iib);
 
 	zpci_ibv = kcalloc(num_possible_cpus(), sizeof(*zpci_ibv),
 			   GFP_KERNEL);
@@ -471,6 +468,7 @@ static int __init zpci_floating_irq_init(void)
 
 int __init zpci_irq_init(void)
 {
+	union zpci_sic_iib iib = {{0}};
 	int rc;
 
 	irq_delivery = sclp.has_dirq ? DIRECTED : FLOATING;
@@ -502,7 +500,7 @@ int __init zpci_irq_init(void)
 	 * Enable floating IRQs (with suppression after one IRQ). When using
 	 * directed IRQs this enables the fallback path.
 	 */
-	zpci_set_irq_ctrl(SIC_IRQ_MODE_SINGLE, PCI_ISC);
+	zpci_set_irq_ctrl(SIC_IRQ_MODE_SINGLE, PCI_ISC, &iib);
 
 	return 0;
 out_airq:
Signature · 1 line
2.27.0
maintainer Reviewed-by on v2 7/30
#13 lore raw

Re: [PATCH v2 07/30] s390/pci: externalize the SIC operation controls and routine

Quoted 7 lines · Matthew Rosato

A subsequent patch will be issuing SIC from KVM -- export the necessary routine and make the operation control definitions available from a header. Because the routine will now be exported, let's rename __zpci_set_irq_ctrl to zpci_set_irq_ctrl and get rid of the zero'd iib wrapper function of the same name.

Looks good thank you!

Quoted 7 lines
---
 arch/s390/include/asm/pci_insn.h | 17 +++++++++--------
 arch/s390/pci/pci_insn.c         |  3 ++-
 arch/s390/pci/pci_irq.c          | 26 ++++++++++++--------------
 3 files changed, 23 insertions(+), 23 deletions(-)

---8<---

maintainer Reviewed-by on v2 7/30
#14 lore raw

Re: [PATCH v2 07/30] s390/pci: externalize the SIC operation controls and routine

In reply to Matthew Rosato
Quoted 7 lines · On Fri, 14 Jan 2022 15:31:22 -0500 Matthew Rosato <[email protected]>…

A subsequent patch will be issuing SIC from KVM -- export the necessary routine and make the operation control definitions available from a header. Because the routine will now be exported, let's rename __zpci_set_irq_ctrl to zpci_set_irq_ctrl and get rid of the zero'd iib wrapper function of the same name.

Quoted 171 lines
---
 arch/s390/include/asm/pci_insn.h | 17 +++++++++--------
 arch/s390/pci/pci_insn.c         |  3 ++-
 arch/s390/pci/pci_irq.c          | 26 ++++++++++++--------------
 3 files changed, 23 insertions(+), 23 deletions(-)
arch/s390/include/asm/pci_insn.h +9 -8
index 61cf9531f68f..5331082fa516 100644
@@ -98,6 +98,14 @@ struct zpci_fib {
 	u32 gd;
 } __packed __aligned(8);
 
+/* Set Interruption Controls Operation Controls  */
+#define	SIC_IRQ_MODE_ALL		0
+#define	SIC_IRQ_MODE_SINGLE		1
+#define	SIC_IRQ_MODE_DIRECT		4
+#define	SIC_IRQ_MODE_D_ALL		16
+#define	SIC_IRQ_MODE_D_SINGLE		17
+#define	SIC_IRQ_MODE_SET_CPU		18
+
 /* directed interruption information block */
 struct zpci_diib {
 	u32 : 1;
@@ -134,13 +142,6 @@ int __zpci_store(u64 data, u64 req, u64 offset);
 int zpci_store(const volatile void __iomem *addr, u64 data, unsigned long len);
 int __zpci_store_block(const u64 *data, u64 req, u64 offset);
 void zpci_barrier(void);
-int __zpci_set_irq_ctrl(u16 ctl, u8 isc, union zpci_sic_iib *iib);
-
-static inline int zpci_set_irq_ctrl(u16 ctl, u8 isc)
-{
-	union zpci_sic_iib iib = {{0}};
-
-	return __zpci_set_irq_ctrl(ctl, isc, &iib);
-}
+int zpci_set_irq_ctrl(u16 ctl, u8 isc, union zpci_sic_iib *iib);
 
 #endif
arch/s390/pci/pci_insn.c +2 -1
index 4dd58b196cea..2a47b3936e44 100644
@@ -97,7 +97,7 @@ int zpci_refresh_trans(u64 fn, u64 addr, u64 range)
 }
 
 /* Set Interruption Controls */
-int __zpci_set_irq_ctrl(u16 ctl, u8 isc, union zpci_sic_iib *iib)
+int zpci_set_irq_ctrl(u16 ctl, u8 isc, union zpci_sic_iib *iib)
 {
 	if (!test_facility(72))
 		return -EIO;
@@ -108,6 +108,7 @@ int __zpci_set_irq_ctrl(u16 ctl, u8 isc, union zpci_sic_iib *iib)
 
 	return 0;
 }
+EXPORT_SYMBOL_GPL(zpci_set_irq_ctrl);
 
 /* PCI Load */
 static inline int ____pcilg(u64 *data, u64 req, u64 offset, u8 *status)
arch/s390/pci/pci_irq.c +12 -14
index 0d0a02a9fbbf..2f675355fd0c 100644
@@ -15,13 +15,6 @@
 
 static enum {FLOATING, DIRECTED} irq_delivery;
 
-#define	SIC_IRQ_MODE_ALL		0
-#define	SIC_IRQ_MODE_SINGLE		1
-#define	SIC_IRQ_MODE_DIRECT		4
-#define	SIC_IRQ_MODE_D_ALL		16
-#define	SIC_IRQ_MODE_D_SINGLE		17
-#define	SIC_IRQ_MODE_SET_CPU		18
-
 /*
  * summary bit vector
  * FLOATING - summary bit per function
@@ -154,6 +147,7 @@ static struct irq_chip zpci_irq_chip = {
 static void zpci_handle_cpu_local_irq(bool rescan)
 {
 	struct airq_iv *dibv = zpci_ibv[smp_processor_id()];
+	union zpci_sic_iib iib = {{0}};
 	unsigned long bit;
 	int irqs_on = 0;
 
@@ -165,7 +159,7 @@ static void zpci_handle_cpu_local_irq(bool rescan)
 				/* End of second scan with interrupts on. */
 				break;
 			/* First scan complete, reenable interrupts. */
-			if (zpci_set_irq_ctrl(SIC_IRQ_MODE_D_SINGLE, PCI_ISC))
+			if (zpci_set_irq_ctrl(SIC_IRQ_MODE_D_SINGLE, PCI_ISC, &iib))
 				break;
 			bit = 0;
 			continue;
@@ -193,6 +187,7 @@ static void zpci_handle_remote_irq(void *data)
 static void zpci_handle_fallback_irq(void)
 {
 	struct cpu_irq_data *cpu_data;
+	union zpci_sic_iib iib = {{0}};
 	unsigned long cpu;
 	int irqs_on = 0;
 
@@ -203,7 +198,7 @@ static void zpci_handle_fallback_irq(void)
 				/* End of second scan with interrupts on. */
 				break;
 			/* First scan complete, reenable interrupts. */
-			if (zpci_set_irq_ctrl(SIC_IRQ_MODE_SINGLE, PCI_ISC))
+			if (zpci_set_irq_ctrl(SIC_IRQ_MODE_SINGLE, PCI_ISC, &iib))
 				break;
 			cpu = 0;
 			continue;
@@ -234,6 +229,7 @@ static void zpci_directed_irq_handler(struct airq_struct *airq,
 static void zpci_floating_irq_handler(struct airq_struct *airq,
 				      struct tpi_info *tpi_info)
 {
+	union zpci_sic_iib iib = {{0}};
 	unsigned long si, ai;
 	struct airq_iv *aibv;
 	int irqs_on = 0;
@@ -247,7 +243,7 @@ static void zpci_floating_irq_handler(struct airq_struct *airq,
 				/* End of second scan with interrupts on. */
 				break;
 			/* First scan complete, reenable interrupts. */
-			if (zpci_set_irq_ctrl(SIC_IRQ_MODE_SINGLE, PCI_ISC))
+			if (zpci_set_irq_ctrl(SIC_IRQ_MODE_SINGLE, PCI_ISC, &iib))
 				break;
 			si = 0;
 			continue;
@@ -407,11 +403,12 @@ static struct airq_struct zpci_airq = {
 static void __init cpu_enable_directed_irq(void *unused)
 {
 	union zpci_sic_iib iib = {{0}};
+	union zpci_sic_iib ziib = {{0}};
 
 	iib.cdiib.dibv_addr = (u64) zpci_ibv[smp_processor_id()]->vector;
 
-	__zpci_set_irq_ctrl(SIC_IRQ_MODE_SET_CPU, 0, &iib);
-	zpci_set_irq_ctrl(SIC_IRQ_MODE_D_SINGLE, PCI_ISC);
+	zpci_set_irq_ctrl(SIC_IRQ_MODE_SET_CPU, 0, &iib);
+	zpci_set_irq_ctrl(SIC_IRQ_MODE_D_SINGLE, PCI_ISC, &ziib);
 }
 
 static int __init zpci_directed_irq_init(void)
@@ -426,7 +423,7 @@ static int __init zpci_directed_irq_init(void)
 	iib.diib.isc = PCI_ISC;
 	iib.diib.nr_cpus = num_possible_cpus();
 	iib.diib.disb_addr = virt_to_phys(zpci_sbv->vector);
-	__zpci_set_irq_ctrl(SIC_IRQ_MODE_DIRECT, 0, &iib);
+	zpci_set_irq_ctrl(SIC_IRQ_MODE_DIRECT, 0, &iib);
 
 	zpci_ibv = kcalloc(num_possible_cpus(), sizeof(*zpci_ibv),
 			   GFP_KERNEL);
@@ -471,6 +468,7 @@ static int __init zpci_floating_irq_init(void)
 
 int __init zpci_irq_init(void)
 {
+	union zpci_sic_iib iib = {{0}};
 	int rc;
 
 	irq_delivery = sclp.has_dirq ? DIRECTED : FLOATING;
@@ -502,7 +500,7 @@ int __init zpci_irq_init(void)
 	 * Enable floating IRQs (with suppression after one IRQ). When using
 	 * directed IRQs this enables the fallback path.
 	 */
-	zpci_set_irq_ctrl(SIC_IRQ_MODE_SINGLE, PCI_ISC);
+	zpci_set_irq_ctrl(SIC_IRQ_MODE_SINGLE, PCI_ISC, &iib);
 
 	return 0;
 out_airq:
Reviewed-by on v2 7/30
#15 lore raw

Re: [PATCH v2 07/30] s390/pci: externalize the SIC operation controls and routine

In reply to Matthew Rosato
Quoted 7 lines · Matthew Rosato

A subsequent patch will be issuing SIC from KVM -- export the necessary routine and make the operation control definitions available from a header. Because the routine will now be exported, let's rename __zpci_set_irq_ctrl to zpci_set_irq_ctrl and get rid of the zero'd iib wrapper function of the same name.

Quoted 175 lines
---
  arch/s390/include/asm/pci_insn.h | 17 +++++++++--------
  arch/s390/pci/pci_insn.c         |  3 ++-
  arch/s390/pci/pci_irq.c          | 26 ++++++++++++--------------
  3 files changed, 23 insertions(+), 23 deletions(-)
arch/s390/include/asm/pci_insn.h +9 -8
index 61cf9531f68f..5331082fa516 100644
@@ -98,6 +98,14 @@ struct zpci_fib {
  	u32 gd;
  } __packed __aligned(8);
 
+/* Set Interruption Controls Operation Controls  */
+#define	SIC_IRQ_MODE_ALL		0
+#define	SIC_IRQ_MODE_SINGLE		1
+#define	SIC_IRQ_MODE_DIRECT		4
+#define	SIC_IRQ_MODE_D_ALL		16
+#define	SIC_IRQ_MODE_D_SINGLE		17
+#define	SIC_IRQ_MODE_SET_CPU		18
+
  /* directed interruption information block */
  struct zpci_diib {
  	u32 : 1;
@@ -134,13 +142,6 @@ int __zpci_store(u64 data, u64 req, u64 offset);
  int zpci_store(const volatile void __iomem *addr, u64 data, unsigned long len);
  int __zpci_store_block(const u64 *data, u64 req, u64 offset);
  void zpci_barrier(void);
-int __zpci_set_irq_ctrl(u16 ctl, u8 isc, union zpci_sic_iib *iib);
-
-static inline int zpci_set_irq_ctrl(u16 ctl, u8 isc)
-{
-	union zpci_sic_iib iib = {{0}};
-
-	return __zpci_set_irq_ctrl(ctl, isc, &iib);
-}
+int zpci_set_irq_ctrl(u16 ctl, u8 isc, union zpci_sic_iib *iib);
 
  #endif
arch/s390/pci/pci_insn.c +2 -1
index 4dd58b196cea..2a47b3936e44 100644
@@ -97,7 +97,7 @@ int zpci_refresh_trans(u64 fn, u64 addr, u64 range)
  }
 
  /* Set Interruption Controls */
-int __zpci_set_irq_ctrl(u16 ctl, u8 isc, union zpci_sic_iib *iib)
+int zpci_set_irq_ctrl(u16 ctl, u8 isc, union zpci_sic_iib *iib)
  {
  	if (!test_facility(72))
  		return -EIO;
@@ -108,6 +108,7 @@ int __zpci_set_irq_ctrl(u16 ctl, u8 isc, union zpci_sic_iib *iib)
 
  	return 0;
  }
+EXPORT_SYMBOL_GPL(zpci_set_irq_ctrl);
 
  /* PCI Load */
  static inline int ____pcilg(u64 *data, u64 req, u64 offset, u8 *status)
arch/s390/pci/pci_irq.c +1 -7
index 0d0a02a9fbbf..2f675355fd0c 100644
@@ -15,13 +15,6 @@
 
  static enum {FLOATING, DIRECTED} irq_delivery;
 
-#define	SIC_IRQ_MODE_ALL		0
-#define	SIC_IRQ_MODE_SINGLE		1
-#define	SIC_IRQ_MODE_DIRECT		4
-#define	SIC_IRQ_MODE_D_ALL		16
-#define	SIC_IRQ_MODE_D_SINGLE		17
-#define	SIC_IRQ_MODE_SET_CPU		18
-
  /*
   * summary bit vector
   * FLOATING - summary bit per function
@@ -154,6 +147,7 @@ static struct irq_chip zpci_irq_chip = {
  static void zpci_handle_cpu_local_irq(bool rescan)
  {
  	struct airq_iv *dibv = zpci_ibv[smp_processor_id()];
+	union zpci_sic_iib iib = {{0}};
 
 
 
  	unsigned long bit;
  	int irqs_on = 0;
arch/s390/pci/pci_irq.c +11 -7
@@ -165,7 +159,7 @@ static void zpci_handle_cpu_local_irq(bool rescan)
  				/* End of second scan with interrupts on. */
  				break;
  			/* First scan complete, reenable interrupts. */
-			if (zpci_set_irq_ctrl(SIC_IRQ_MODE_D_SINGLE, PCI_ISC))
+			if (zpci_set_irq_ctrl(SIC_IRQ_MODE_D_SINGLE, PCI_ISC, &iib))
  				break;
  			bit = 0;
  			continue;
@@ -193,6 +187,7 @@ static void zpci_handle_remote_irq(void *data)
  static void zpci_handle_fallback_irq(void)
  {
  	struct cpu_irq_data *cpu_data;
+	union zpci_sic_iib iib = {{0}};
  	unsigned long cpu;
  	int irqs_on = 0;
 
@@ -203,7 +198,7 @@ static void zpci_handle_fallback_irq(void)
  				/* End of second scan with interrupts on. */
  				break;
  			/* First scan complete, reenable interrupts. */
-			if (zpci_set_irq_ctrl(SIC_IRQ_MODE_SINGLE, PCI_ISC))
+			if (zpci_set_irq_ctrl(SIC_IRQ_MODE_SINGLE, PCI_ISC, &iib))
  				break;
  			cpu = 0;
  			continue;
@@ -234,6 +229,7 @@ static void zpci_directed_irq_handler(struct airq_struct *airq,
  static void zpci_floating_irq_handler(struct airq_struct *airq,
  				      struct tpi_info *tpi_info)
  {
+	union zpci_sic_iib iib = {{0}};
  	unsigned long si, ai;
  	struct airq_iv *aibv;
  	int irqs_on = 0;
@@ -247,7 +243,7 @@ static void zpci_floating_irq_handler(struct airq_struct *airq,
  				/* End of second scan with interrupts on. */
  				break;
  			/* First scan complete, reenable interrupts. */
-			if (zpci_set_irq_ctrl(SIC_IRQ_MODE_SINGLE, PCI_ISC))
+			if (zpci_set_irq_ctrl(SIC_IRQ_MODE_SINGLE, PCI_ISC, &iib))
  				break;
  			si = 0;
  			continue;
@@ -407,11 +403,12 @@ static struct airq_struct zpci_airq = {
  static void __init cpu_enable_directed_irq(void *unused)
  {
  	union zpci_sic_iib iib = {{0}};
+	union zpci_sic_iib ziib = {{0}};
 
  	iib.cdiib.dibv_addr = (u64) zpci_ibv[smp_processor_id()]->vector;
 
-	__zpci_set_irq_ctrl(SIC_IRQ_MODE_SET_CPU, 0, &iib);
-	zpci_set_irq_ctrl(SIC_IRQ_MODE_D_SINGLE, PCI_ISC);
+	zpci_set_irq_ctrl(SIC_IRQ_MODE_SET_CPU, 0, &iib);
+	zpci_set_irq_ctrl(SIC_IRQ_MODE_D_SINGLE, PCI_ISC, &ziib);
  }
 
  static int __init zpci_directed_irq_init(void)
@@ -426,7 +423,7 @@ static int __init zpci_directed_irq_init(void)
  	iib.diib.isc = PCI_ISC;
  	iib.diib.nr_cpus = num_possible_cpus();
  	iib.diib.disb_addr = virt_to_phys(zpci_sbv->vector);
-	__zpci_set_irq_ctrl(SIC_IRQ_MODE_DIRECT, 0, &iib);
+	zpci_set_irq_ctrl(SIC_IRQ_MODE_DIRECT, 0, &iib);
 
  	zpci_ibv = kcalloc(num_possible_cpus(), sizeof(*zpci_ibv),
  			   GFP_KERNEL);
@@ -471,6 +468,7 @@ static int __init zpci_floating_irq_init(void)
 
  int __init zpci_irq_init(void)
  {
+	union zpci_sic_iib iib = {{0}};
  	int rc;
 
  	irq_delivery = sclp.has_dirq ? DIRECTED : FLOATING;
@@ -502,7 +500,7 @@ int __init zpci_irq_init(void)
  	 * Enable floating IRQs (with suppression after one IRQ). When using
  	 * directed IRQs this enables the fallback path.
  	 */
-	zpci_set_irq_ctrl(SIC_IRQ_MODE_SINGLE, PCI_ISC);
+	zpci_set_irq_ctrl(SIC_IRQ_MODE_SINGLE, PCI_ISC, &iib);
 
  	return 0;
  out_airq:
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#16 lore raw

PATCH v2 8/30 s390/pci: stash associated GISA designation

In reply to Matthew Rosato

For passthrough devices, we will need to know the GISA designation of the guest if interpretation facilities are to be used. Setup to stash this in the zdev and set a default of 0 (no GISA designation) for now; a subsequent patch will set a valid GISA designation for passthrough devices. Also, extend mpcific routines to specify this stashed designation as part of the mpcific command.

---
 arch/s390/include/asm/pci.h     | 1 +
 arch/s390/include/asm/pci_clp.h | 3 ++-
 arch/s390/pci/pci.c             | 6 ++++++
 arch/s390/pci/pci_clp.c         | 1 +
 arch/s390/pci/pci_irq.c         | 5 +++++
 5 files changed, 15 insertions(+), 1 deletion(-)
arch/s390/include/asm/pci.h +1 -0
index 90824be5ce9a..2474b8d30f2a 100644
@@ -123,6 +123,7 @@ struct zpci_dev {
 	enum zpci_state state;
 	u32		fid;		/* function ID, used by sclp */
 	u32		fh;		/* function handle, used by insn's */
+	u32		gd;		/* GISA designation for passthrough */
 	u16		vfn;		/* virtual function number */
 	u16		pchid;		/* physical channel ID */
 	u8		pfgid;		/* function group ID */
arch/s390/include/asm/pci_clp.h +2 -1
index 1f4b666e85ee..3af8d196da74 100644
@@ -173,7 +173,8 @@ struct clp_req_set_pci {
 	u16 reserved2;
 	u8 oc;				/* operation controls */
 	u8 ndas;			/* number of dma spaces */
-	u64 reserved3;
+	u32 reserved3;
+	u32 gd;				/* GISA designation */
 } __packed;
 
 /* Set PCI function response */
arch/s390/pci/pci.c +6 -0
index 792f8e0f2178..0c9879dae752 100644
@@ -119,6 +119,7 @@ int zpci_register_ioat(struct zpci_dev *zdev, u8 dmaas,
 	fib.pba = base;
 	fib.pal = limit;
 	fib.iota = iota | ZPCI_IOTA_RTTO_FLAG;
+	fib.gd = zdev->gd;
 	cc = zpci_mod_fc(req, &fib, &status);
 	if (cc)
 		zpci_dbg(3, "reg ioat fid:%x, cc:%d, status:%d\n", zdev->fid, cc, status);
@@ -132,6 +133,8 @@ int zpci_unregister_ioat(struct zpci_dev *zdev, u8 dmaas)
 	struct zpci_fib fib = {0};
 	u8 cc, status;
 
+	fib.gd = zdev->gd;
+
 	cc = zpci_mod_fc(req, &fib, &status);
 	if (cc)
 		zpci_dbg(3, "unreg ioat fid:%x, cc:%d, status:%d\n", zdev->fid, cc, status);
@@ -159,6 +162,7 @@ int zpci_fmb_enable_device(struct zpci_dev *zdev)
 	atomic64_set(&zdev->unmapped_pages, 0);
 
 	fib.fmb_addr = virt_to_phys(zdev->fmb);
+	fib.gd = zdev->gd;
 	cc = zpci_mod_fc(req, &fib, &status);
 	if (cc) {
 		kmem_cache_free(zdev_fmb_cache, zdev->fmb);
@@ -177,6 +181,8 @@ int zpci_fmb_disable_device(struct zpci_dev *zdev)
 	if (!zdev->fmb)
 		return -EINVAL;
 
+	fib.gd = zdev->gd;
+
 	/* Function measurement is disabled if fmb address is zero */
 	cc = zpci_mod_fc(req, &fib, &status);
 	if (cc == 3) /* Function already gone. */
arch/s390/pci/pci_clp.c +1 -0
index be077b39da33..e9ed0e4a5cf0 100644
@@ -240,6 +240,7 @@ static int clp_set_pci_fn(struct zpci_dev *zdev, u32 *fh, u8 nr_dma_as, u8 comma
 		rrb->request.fh = zdev->fh;
 		rrb->request.oc = command;
 		rrb->request.ndas = nr_dma_as;
+		rrb->request.gd = zdev->gd;
 
 		rc = clp_req(rrb, CLP_LPS_PCI);
 		if (rrb->response.hdr.rsp == CLP_RC_SETPCIFN_BUSY) {
arch/s390/pci/pci_irq.c +5 -0
index 2f675355fd0c..17e5adfe1273 100644
@@ -43,6 +43,7 @@ static int zpci_set_airq(struct zpci_dev *zdev)
 	fib.fmt0.aibvo = 0;	/* each zdev has its own interrupt vector */
 	fib.fmt0.aisb = virt_to_phys(zpci_sbv->vector) + (zdev->aisb / 64) * 8;
 	fib.fmt0.aisbo = zdev->aisb & 63;
+	fib.gd = zdev->gd;
 
 	return zpci_mod_fc(req, &fib, &status) ? -EIO : 0;
 }
@@ -54,6 +55,8 @@ static int zpci_clear_airq(struct zpci_dev *zdev)
 	struct zpci_fib fib = {0};
 	u8 cc, status;
 
+	fib.gd = zdev->gd;
+
 	cc = zpci_mod_fc(req, &fib, &status);
 	if (cc == 3 || (cc == 1 && status == 24))
 		/* Function already gone or IRQs already deregistered. */
@@ -72,6 +75,7 @@ static int zpci_set_directed_irq(struct zpci_dev *zdev)
 	fib.fmt = 1;
 	fib.fmt1.noi = zdev->msi_nr_irqs;
 	fib.fmt1.dibvo = zdev->msi_first_bit;
+	fib.gd = zdev->gd;
 
 	return zpci_mod_fc(req, &fib, &status) ? -EIO : 0;
 }
@@ -84,6 +88,7 @@ static int zpci_clear_directed_irq(struct zpci_dev *zdev)
 	u8 cc, status;
 
 	fib.fmt = 1;
+	fib.gd = zdev->gd;
 	cc = zpci_mod_fc(req, &fib, &status);
 	if (cc == 3 || (cc == 1 && status == 24))
 		/* Function already gone or IRQs already deregistered. */
Signature · 1 line
2.27.0
#17 lore raw

Re: [PATCH v2 08/30] s390/pci: stash associated GISA designation

Quoted 25 lines · Matthew Rosato

For passthrough devices, we will need to know the GISA designation of the guest if interpretation facilities are to be used. Setup to stash this in the zdev and set a default of 0 (no GISA designation) for now; a subsequent patch will set a valid GISA designation for passthrough devices. Also, extend mpcific routines to specify this stashed designation as part of the mpcific command.

---
  arch/s390/include/asm/pci.h     | 1 +
  arch/s390/include/asm/pci_clp.h | 3 ++-
  arch/s390/pci/pci.c             | 6 ++++++
  arch/s390/pci/pci_clp.c         | 1 +
  arch/s390/pci/pci_irq.c         | 5 +++++
  5 files changed, 15 insertions(+), 1 deletion(-)
arch/s390/include/asm/pci.h +0 -0
index 90824be5ce9a..2474b8d30f2a 100644

arch/s390/include/asm/pci.h struct zpci_dev { · line 126

Quoted 4 lines
arch/s390/include/asm/pci.h +1 -0
@@ -123,6 +123,7 @@ struct zpci_dev {
  	enum zpci_state state;
  	u32		fid;		/* function ID, used by sclp */
  	u32		fh;		/* function handle, used by insn's */
+	u32		gd;		/* GISA designation for passthrough */

I already gave my R-B, and do not want to remove it, but wouldn't it be possible to use more explicit names like gisa_designation instead of just gd. It would not change anything to the functionality but would facilitate the maintenance?

Quoted 8 lines
arch/s390/include/asm/pci.h +0 -0
  	u16		vfn;		/* virtual function number */
  	u16		pchid;		/* physical channel ID */
  	u8		pfgid;		/* function group ID */
arch/s390/include/asm/pci_clp.h +0 -0
index 1f4b666e85ee..3af8d196da74 100644

arch/s390/include/asm/pci_clp.h struct clp_req_set_pci { · line 177

Quoted 6 lines
arch/s390/include/asm/pci_clp.h +2 -1
@@ -173,7 +173,8 @@ struct clp_req_set_pci {
  	u16 reserved2;
  	u8 oc;				/* operation controls */
  	u8 ndas;			/* number of dma spaces */
-	u64 reserved3;
+	u32 reserved3;
+	u32 gd;				/* GISA designation */

here too.

Quoted 91 lines
arch/s390/include/asm/pci_clp.h +0 -0
  } __packed;
 
  /* Set PCI function response */
arch/s390/pci/pci.c +6 -0
index 792f8e0f2178..0c9879dae752 100644
@@ -119,6 +119,7 @@ int zpci_register_ioat(struct zpci_dev *zdev, u8 dmaas,
  	fib.pba = base;
  	fib.pal = limit;
  	fib.iota = iota | ZPCI_IOTA_RTTO_FLAG;
+	fib.gd = zdev->gd;
  	cc = zpci_mod_fc(req, &fib, &status);
  	if (cc)
  		zpci_dbg(3, "reg ioat fid:%x, cc:%d, status:%d\n", zdev->fid, cc, status);
@@ -132,6 +133,8 @@ int zpci_unregister_ioat(struct zpci_dev *zdev, u8 dmaas)
  	struct zpci_fib fib = {0};
  	u8 cc, status;
 
+	fib.gd = zdev->gd;
+
  	cc = zpci_mod_fc(req, &fib, &status);
  	if (cc)
  		zpci_dbg(3, "unreg ioat fid:%x, cc:%d, status:%d\n", zdev->fid, cc, status);
@@ -159,6 +162,7 @@ int zpci_fmb_enable_device(struct zpci_dev *zdev)
  	atomic64_set(&zdev->unmapped_pages, 0);
 
  	fib.fmb_addr = virt_to_phys(zdev->fmb);
+	fib.gd = zdev->gd;
  	cc = zpci_mod_fc(req, &fib, &status);
  	if (cc) {
  		kmem_cache_free(zdev_fmb_cache, zdev->fmb);
@@ -177,6 +181,8 @@ int zpci_fmb_disable_device(struct zpci_dev *zdev)
  	if (!zdev->fmb)
  		return -EINVAL;
 
+	fib.gd = zdev->gd;
+
  	/* Function measurement is disabled if fmb address is zero */
  	cc = zpci_mod_fc(req, &fib, &status);
  	if (cc == 3) /* Function already gone. */
arch/s390/pci/pci_clp.c +1 -0
index be077b39da33..e9ed0e4a5cf0 100644
@@ -240,6 +240,7 @@ static int clp_set_pci_fn(struct zpci_dev *zdev, u32 *fh, u8 nr_dma_as, u8 comma
  		rrb->request.fh = zdev->fh;
  		rrb->request.oc = command;
  		rrb->request.ndas = nr_dma_as;
+		rrb->request.gd = zdev->gd;
 
  		rc = clp_req(rrb, CLP_LPS_PCI);
  		if (rrb->response.hdr.rsp == CLP_RC_SETPCIFN_BUSY) {
arch/s390/pci/pci_irq.c +5 -0
index 2f675355fd0c..17e5adfe1273 100644
@@ -43,6 +43,7 @@ static int zpci_set_airq(struct zpci_dev *zdev)
  	fib.fmt0.aibvo = 0;	/* each zdev has its own interrupt vector */
  	fib.fmt0.aisb = virt_to_phys(zpci_sbv->vector) + (zdev->aisb / 64) * 8;
  	fib.fmt0.aisbo = zdev->aisb & 63;
+	fib.gd = zdev->gd;
 
  	return zpci_mod_fc(req, &fib, &status) ? -EIO : 0;
  }
@@ -54,6 +55,8 @@ static int zpci_clear_airq(struct zpci_dev *zdev)
  	struct zpci_fib fib = {0};
  	u8 cc, status;
 
+	fib.gd = zdev->gd;
+
  	cc = zpci_mod_fc(req, &fib, &status);
  	if (cc == 3 || (cc == 1 && status == 24))
  		/* Function already gone or IRQs already deregistered. */
@@ -72,6 +75,7 @@ static int zpci_set_directed_irq(struct zpci_dev *zdev)
  	fib.fmt = 1;
  	fib.fmt1.noi = zdev->msi_nr_irqs;
  	fib.fmt1.dibvo = zdev->msi_first_bit;
+	fib.gd = zdev->gd;
 
  	return zpci_mod_fc(req, &fib, &status) ? -EIO : 0;
  }
@@ -84,6 +88,7 @@ static int zpci_clear_directed_irq(struct zpci_dev *zdev)
  	u8 cc, status;
 
  	fib.fmt = 1;
+	fib.gd = zdev->gd;
  	cc = zpci_mod_fc(req, &fib, &status);
  	if (cc == 3 || (cc == 1 && status == 24))
  		/* Function already gone or IRQs already deregistered. */
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#18 lore raw

Re: [PATCH v2 08/30] s390/pci: stash associated GISA designation

Quoted 23 lines · Pierre Morel
Quoted 20 lines · Matthew Rosato

For passthrough devices, we will need to know the GISA designation of the guest if interpretation facilities are to be used.  Setup to stash this in the zdev and set a default of 0 (no GISA designation) for now; a subsequent patch will set a valid GISA designation for passthrough devices. Also, extend mpcific routines to specify this stashed designation as part of the mpcific command.

---
  arch/s390/include/asm/pci.h     | 1 +
  arch/s390/include/asm/pci_clp.h | 3 ++-
  arch/s390/pci/pci.c             | 6 ++++++
  arch/s390/pci/pci_clp.c         | 1 +
  arch/s390/pci/pci_irq.c         | 5 +++++

  5 files changed, 15 insertions(+), 1 deletion(-)
Quoted 13 lines
Quoted 9 lines
arch/s390/include/asm/pci.h +0 -0
index 90824be5ce9a..2474b8d30f2a 100644
@@ -123,6 +123,7 @@ struct zpci_dev {
       enum zpci_state state;

      u32        fid;        /* function ID, used by sclp */ u32        fh;        /* function handle, used by insn's */ +    u32        gd;        /* GISA designation for passthrough */

I already gave my R-B, and do not want to remove it, but wouldn't it be possible to use more explicit names like gisa_designation instead of just gd. It would not change anything to the functionality but would facilitate the maintenance?

Honestly, I don't have a strong opinion on this one -- AFAICT struct zpci_dev has a fair mix of short names (fh) and explicit names (max_bus_speed).

It does require changes to this patch and various subsequent patches -- The changes are, as you say, not functional, so I think it's not a big deal?

I do think 'gisa_designation' is too verbose though -- How about just 'gisa', this is the same name used in the structure where we get this value from (gisa in struct sie_page2)

As long as nobody objects I will s/gd/gisa/ here and in struct clp_req_set_pci, retaining review tags.

author maintainer
#19 lore raw

PATCH v2 9/30 s390/pci: export some routines related to RPCIT processing

In reply to Matthew Rosato

KVM will re-use dma_walk_cpu_trans to walk the host shadow table and will also need to be able to call zpci_refresh_trans to re-issue a RPCIT.

---
 arch/s390/pci/pci_dma.c  | 1 +
 arch/s390/pci/pci_insn.c | 1 +
 2 files changed, 2 insertions(+)
arch/s390/pci/pci_dma.c +1 -0
index f46833a25526..a81de48d5ea7 100644
@@ -116,6 +116,7 @@ unsigned long *dma_walk_cpu_trans(unsigned long *rto, dma_addr_t dma_addr)
 	px = calc_px(dma_addr);
 	return &pto[px];
 }
+EXPORT_SYMBOL_GPL(dma_walk_cpu_trans);
 
 void dma_update_cpu_trans(unsigned long *entry, phys_addr_t page_addr, int flags)
 {
arch/s390/pci/pci_insn.c +1 -0
index 2a47b3936e44..0509554301c7 100644
@@ -95,6 +95,7 @@ int zpci_refresh_trans(u64 fn, u64 addr, u64 range)
 
 	return (cc) ? -EIO : 0;
 }
+EXPORT_SYMBOL_GPL(zpci_refresh_trans);
 
 /* Set Interruption Controls */
 int zpci_set_irq_ctrl(u16 ctl, u8 isc, union zpci_sic_iib *iib)
Signature · 1 line
2.27.0
Reviewed-by on v2 9/30
#20 lore raw

Re: [PATCH v2 09/30] s390/pci: export some routines related to RPCIT processing

Quoted 6 lines · Matthew Rosato

KVM will re-use dma_walk_cpu_trans to walk the host shadow table and will also need to be able to call zpci_refresh_trans to re-issue a RPCIT.

Quoted 30 lines
---
  arch/s390/pci/pci_dma.c  | 1 +
  arch/s390/pci/pci_insn.c | 1 +
  2 files changed, 2 insertions(+)
arch/s390/pci/pci_dma.c +1 -0
index f46833a25526..a81de48d5ea7 100644
@@ -116,6 +116,7 @@ unsigned long *dma_walk_cpu_trans(unsigned long *rto, dma_addr_t dma_addr)
  	px = calc_px(dma_addr);
  	return &pto[px];
  }
+EXPORT_SYMBOL_GPL(dma_walk_cpu_trans);
 
  void dma_update_cpu_trans(unsigned long *entry, phys_addr_t page_addr, int flags)
  {
arch/s390/pci/pci_insn.c +1 -0
index 2a47b3936e44..0509554301c7 100644
@@ -95,6 +95,7 @@ int zpci_refresh_trans(u64 fn, u64 addr, u64 range)
 
  	return (cc) ? -EIO : 0;
  }
+EXPORT_SYMBOL_GPL(zpci_refresh_trans);
 
  /* Set Interruption Controls */
  int zpci_set_irq_ctrl(u16 ctl, u8 isc, union zpci_sic_iib *iib)
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#21 lore raw

PATCH v2 10/30 s390/pci: stash dtsm and maxstbl

In reply to Matthew Rosato

Store information about what IOAT designation types are supported by underlying hardware as well as the largest store block size allowed. These values will be needed by passthrough.

---
 arch/s390/include/asm/pci.h     | 2 ++
 arch/s390/include/asm/pci_clp.h | 6 ++++--
 arch/s390/pci/pci_clp.c         | 2 ++
 3 files changed, 8 insertions(+), 2 deletions(-)
arch/s390/include/asm/pci.h +2 -0
index 2474b8d30f2a..1a8f9f42da3a 100644
@@ -126,9 +126,11 @@ struct zpci_dev {
 	u32		gd;		/* GISA designation for passthrough */
 	u16		vfn;		/* virtual function number */
 	u16		pchid;		/* physical channel ID */
+	u16		maxstbl;	/* Maximum store block size */
 	u8		pfgid;		/* function group ID */
 	u8		pft;		/* pci function type */
 	u8		port;
+	u8		dtsm;		/* Supported DT mask */
 	u8		rid_available	: 1;
 	u8		has_hp_slot	: 1;
 	u8		has_resources	: 1;
arch/s390/include/asm/pci_clp.h +4 -2
index 3af8d196da74..124fadfb74b9 100644
@@ -153,9 +153,11 @@ struct clp_rsp_query_pci_grp {
 	u8			:  6;
 	u8 frame		:  1;
 	u8 refresh		:  1;	/* TLB refresh mode */
-	u16 reserved2;
+	u16			:  3;
+	u16 maxstbl		: 13;	/* Maximum store block size */
 	u16 mui;
-	u16			: 16;
+	u8 dtsm;			/* Supported DT mask */
+	u8 reserved3;
 	u16 maxfaal;
 	u16			:  4;
 	u16 dnoi		: 12;
arch/s390/pci/pci_clp.c +2 -0
index e9ed0e4a5cf0..bc7446566cbc 100644
@@ -103,6 +103,8 @@ static void clp_store_query_pci_fngrp(struct zpci_dev *zdev,
 	zdev->max_msi = response->noi;
 	zdev->fmb_update = response->mui;
 	zdev->version = response->version;
+	zdev->maxstbl = response->maxstbl;
+	zdev->dtsm = response->dtsm;
 
 	switch (response->version) {
 	case 1:
Signature · 1 line
2.27.0
author maintainer
#22 lore raw

PATCH v2 11/30 s390/pci: add helper function to find device by handle

In reply to Matthew Rosato

Intercepted zPCI instructions will specify the desired function via a function handle. Add a routine to find the device with the specified handle.

---
 arch/s390/include/asm/pci.h |  1 +
 arch/s390/pci/pci.c         | 16 ++++++++++++++++
 2 files changed, 17 insertions(+)
arch/s390/include/asm/pci.h +1 -0
index 1a8f9f42da3a..00a2c24d6d2b 100644
@@ -275,6 +275,7 @@ static inline struct zpci_dev *to_zpci_dev(struct device *dev)
 }
 
 struct zpci_dev *get_zdev_by_fid(u32);
+struct zpci_dev *get_zdev_by_fh(u32 fh);
 
 /* DMA */
 int zpci_dma_init(void);
arch/s390/pci/pci.c +16 -0
index 0c9879dae752..1e939b4cf25e 100644
@@ -76,6 +76,22 @@ struct zpci_dev *get_zdev_by_fid(u32 fid)
 	return zdev;
 }
 
+struct zpci_dev *get_zdev_by_fh(u32 fh)
+{
+	struct zpci_dev *tmp, *zdev = NULL;
+
+	spin_lock(&zpci_list_lock);
+	list_for_each_entry(tmp, &zpci_list, entry) {
+		if (tmp->fh == fh) {
+			zdev = tmp;
+			break;
+		}
+	}
+	spin_unlock(&zpci_list_lock);
+	return zdev;
+}
+EXPORT_SYMBOL_GPL(get_zdev_by_fh);
+
 void zpci_remove_reserved_devices(void)
 {
 	struct zpci_dev *tmp, *zdev;
Signature · 1 line
2.27.0
Reviewed-by on v2 11/30
#23 lore raw

Re: [PATCH v2 11/30] s390/pci: add helper function to find device by handle

Quoted 8 lines · Matthew Rosato

Intercepted zPCI instructions will specify the desired function via a function handle. Add a routine to find the device with the specified handle.

Quoted 45 lines
---
  arch/s390/include/asm/pci.h |  1 +
  arch/s390/pci/pci.c         | 16 ++++++++++++++++
  2 files changed, 17 insertions(+)
arch/s390/include/asm/pci.h +1 -0
index 1a8f9f42da3a..00a2c24d6d2b 100644
@@ -275,6 +275,7 @@ static inline struct zpci_dev *to_zpci_dev(struct device *dev)
  }
 
  struct zpci_dev *get_zdev_by_fid(u32);
+struct zpci_dev *get_zdev_by_fh(u32 fh);
 
  /* DMA */
  int zpci_dma_init(void);
arch/s390/pci/pci.c +16 -0
index 0c9879dae752..1e939b4cf25e 100644
@@ -76,6 +76,22 @@ struct zpci_dev *get_zdev_by_fid(u32 fid)
  	return zdev;
  }
 
+struct zpci_dev *get_zdev_by_fh(u32 fh)
+{
+	struct zpci_dev *tmp, *zdev = NULL;
+
+	spin_lock(&zpci_list_lock);
+	list_for_each_entry(tmp, &zpci_list, entry) {
+		if (tmp->fh == fh) {
+			zdev = tmp;
+			break;
+		}
+	}
+	spin_unlock(&zpci_list_lock);
+	return zdev;
+}
+EXPORT_SYMBOL_GPL(get_zdev_by_fh);
+
  void zpci_remove_reserved_devices(void)
  {
  	struct zpci_dev *tmp, *zdev;
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#24 lore raw

PATCH v2 12/30 s390/pci: get SHM information from list pci

In reply to Matthew Rosato

KVM will need information on the special handle mask used to indicate emulated devices. In order to obtain this, a new type of list pci call must be made to gather the information. Extend clp_list_pci_req to also fetch the model-dependent-data field that holds this mask.

---
 arch/s390/include/asm/pci.h     |  1 +
 arch/s390/include/asm/pci_clp.h |  2 +-
 arch/s390/pci/pci_clp.c         | 28 +++++++++++++++++++++++++---
 3 files changed, 27 insertions(+), 4 deletions(-)
arch/s390/include/asm/pci.h +1 -0
index 00a2c24d6d2b..f3cd2da8128c 100644
@@ -227,6 +227,7 @@ int clp_enable_fh(struct zpci_dev *zdev, u32 *fh, u8 nr_dma_as);
 int clp_disable_fh(struct zpci_dev *zdev, u32 *fh);
 int clp_get_state(u32 fid, enum zpci_state *state);
 int clp_refresh_fh(u32 fid, u32 *fh);
+int zpci_get_mdd(u32 *mdd);
 
 /* UID */
 void update_uid_checking(bool new);
arch/s390/include/asm/pci_clp.h +1 -1
index 124fadfb74b9..d6bc324763f3 100644
@@ -76,7 +76,7 @@ struct clp_req_list_pci {
 struct clp_rsp_list_pci {
 	struct clp_rsp_hdr hdr;
 	u64 resume_token;
-	u32 reserved2;
+	u32 mdd;
 	u16 max_fn;
 	u8			: 7;
 	u8 uid_checking		: 1;
arch/s390/pci/pci_clp.c +25 -3
index bc7446566cbc..308ffb93413f 100644
@@ -328,7 +328,7 @@ int clp_disable_fh(struct zpci_dev *zdev, u32 *fh)
 }
 
 static int clp_list_pci_req(struct clp_req_rsp_list_pci *rrb,
-			    u64 *resume_token, int *nentries)
+			    u64 *resume_token, int *nentries, u32 *mdd)
 {
 	int rc;
 
@@ -354,6 +354,8 @@ static int clp_list_pci_req(struct clp_req_rsp_list_pci *rrb,
 	*nentries = (rrb->response.hdr.len - LIST_PCI_HDR_LEN) /
 		rrb->response.entry_size;
 	*resume_token = rrb->response.resume_token;
+	if (mdd)
+		*mdd = rrb->response.mdd;
 
 	return rc;
 }
@@ -365,7 +367,7 @@ static int clp_list_pci(struct clp_req_rsp_list_pci *rrb, void *data,
 	int nentries, i, rc;
 
 	do {
-		rc = clp_list_pci_req(rrb, &resume_token, &nentries);
+		rc = clp_list_pci_req(rrb, &resume_token, &nentries, NULL);
 		if (rc)
 			return rc;
 		for (i = 0; i < nentries; i++)
@@ -383,7 +385,7 @@ static int clp_find_pci(struct clp_req_rsp_list_pci *rrb, u32 fid,
 	int nentries, i, rc;
 
 	do {
-		rc = clp_list_pci_req(rrb, &resume_token, &nentries);
+		rc = clp_list_pci_req(rrb, &resume_token, &nentries, NULL);
 		if (rc)
 			return rc;
 		fh_list = rrb->response.fh_list;
@@ -468,6 +470,26 @@ int clp_get_state(u32 fid, enum zpci_state *state)
 	return rc;
 }
 
+int zpci_get_mdd(u32 *mdd)
+{
+	struct clp_req_rsp_list_pci *rrb;
+	u64 resume_token = 0;
+	int nentries, rc;
+
+	if (!mdd)
+		return -EINVAL;
+
+	rrb = clp_alloc_block(GFP_KERNEL);
+	if (!rrb)
+		return -ENOMEM;
+
+	rc = clp_list_pci_req(rrb, &resume_token, &nentries, mdd);
+
+	clp_free_block(rrb);
+	return rc;
+}
+EXPORT_SYMBOL_GPL(zpci_get_mdd);
+
 static int clp_base_slpc(struct clp_req *req, struct clp_req_rsp_slpc *lpcb)
 {
 	unsigned long limit = PAGE_SIZE - sizeof(lpcb->request);
Signature · 1 line
2.27.0
#25 lore raw

Re: [PATCH v2 12/30] s390/pci: get SHM information from list pci

Quoted 79 lines · Matthew Rosato

KVM will need information on the special handle mask used to indicate emulated devices. In order to obtain this, a new type of list pci call must be made to gather the information. Extend clp_list_pci_req to also fetch the model-dependent-data field that holds this mask.

---
  arch/s390/include/asm/pci.h     |  1 +
  arch/s390/include/asm/pci_clp.h |  2 +-
  arch/s390/pci/pci_clp.c         | 28 +++++++++++++++++++++++++---
  3 files changed, 27 insertions(+), 4 deletions(-)
arch/s390/include/asm/pci.h +1 -0
index 00a2c24d6d2b..f3cd2da8128c 100644
@@ -227,6 +227,7 @@ int clp_enable_fh(struct zpci_dev *zdev, u32 *fh, u8 nr_dma_as);
  int clp_disable_fh(struct zpci_dev *zdev, u32 *fh);
  int clp_get_state(u32 fid, enum zpci_state *state);
  int clp_refresh_fh(u32 fid, u32 *fh);
+int zpci_get_mdd(u32 *mdd);
 
  /* UID */
  void update_uid_checking(bool new);
arch/s390/include/asm/pci_clp.h +1 -1
index 124fadfb74b9..d6bc324763f3 100644
@@ -76,7 +76,7 @@ struct clp_req_list_pci {
  struct clp_rsp_list_pci {
  	struct clp_rsp_hdr hdr;
  	u64 resume_token;
-	u32 reserved2;
+	u32 mdd;
  	u16 max_fn;
  	u8			: 7;
  	u8 uid_checking		: 1;
arch/s390/pci/pci_clp.c +5 -3
index bc7446566cbc..308ffb93413f 100644
@@ -328,7 +328,7 @@ int clp_disable_fh(struct zpci_dev *zdev, u32 *fh)
  }
 
  static int clp_list_pci_req(struct clp_req_rsp_list_pci *rrb,
-			    u64 *resume_token, int *nentries)
+			    u64 *resume_token, int *nentries, u32 *mdd)
  {
  	int rc;
 
@@ -354,6 +354,8 @@ static int clp_list_pci_req(struct clp_req_rsp_list_pci *rrb,
  	*nentries = (rrb->response.hdr.len - LIST_PCI_HDR_LEN) /
  		rrb->response.entry_size;
  	*resume_token = rrb->response.resume_token;
+	if (mdd)
+		*mdd = rrb->response.mdd;
 
  	return rc;
  }
@@ -365,7 +367,7 @@ static int clp_list_pci(struct clp_req_rsp_list_pci *rrb, void *data,
  	int nentries, i, rc;
 
  	do {
-		rc = clp_list_pci_req(rrb, &resume_token, &nentries);
+		rc = clp_list_pci_req(rrb, &resume_token, &nentries, NULL);
  		if (rc)
  			return rc;
  		for (i = 0; i < nentries; i++)
@@ -383,7 +385,7 @@ static int clp_find_pci(struct clp_req_rsp_list_pci *rrb, u32 fid,
  	int nentries, i, rc;
 
  	do {
-		rc = clp_list_pci_req(rrb, &resume_token, &nentries);
+		rc = clp_list_pci_req(rrb, &resume_token, &nentries, NULL);
  		if (rc)
  			return rc;
  		fh_list = rrb->response.fh_list;
@@ -468,6 +470,26 @@ int clp_get_state(u32 fid, enum zpci_state *state)
  	return rc;

arch/s390/pci/pci_clp.c int clp_get_state(u32 fid, enum zpci_state *state) · line 480

Quoted 10 lines
arch/s390/pci/pci_clp.c +8 -0
  }
 
+int zpci_get_mdd(u32 *mdd)
+{
+	struct clp_req_rsp_list_pci *rrb;
+	u64 resume_token = 0;
+	int nentries, rc;
+
+	if (!mdd)
+		return -EINVAL;

I think this tests is not useful.
The caller must take care not to call with a NULL pointer, what the only caller today make sure.

Quoted 16 lines
arch/s390/pci/pci_clp.c +12 -0
+
+	rrb = clp_alloc_block(GFP_KERNEL);
+	if (!rrb)
+		return -ENOMEM;
+
+	rc = clp_list_pci_req(rrb, &resume_token, &nentries, mdd);
+
+	clp_free_block(rrb);
+	return rc;
+}
+EXPORT_SYMBOL_GPL(zpci_get_mdd);
+
  static int clp_base_slpc(struct clp_req *req, struct clp_req_rsp_slpc *lpcb)
  {
  	unsigned long limit = PAGE_SIZE - sizeof(lpcb->request);
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
maintainer
#26 lore raw

Re: [PATCH v2 12/30] s390/pci: get SHM information from list pci

Quoted 82 lines · On Tue, 18 Jan 2022 11:36:06 +0100 Pierre Morel <[email protected]> wrote:
Quoted 81 lines · Matthew Rosato

KVM will need information on the special handle mask used to indicate emulated devices. In order to obtain this, a new type of list pci call must be made to gather the information. Extend clp_list_pci_req to also fetch the model-dependent-data field that holds this mask.

---
  arch/s390/include/asm/pci.h     |  1 +
  arch/s390/include/asm/pci_clp.h |  2 +-
  arch/s390/pci/pci_clp.c         | 28 +++++++++++++++++++++++++---
  3 files changed, 27 insertions(+), 4 deletions(-)
arch/s390/include/asm/pci.h +1 -0
index 00a2c24d6d2b..f3cd2da8128c 100644
@@ -227,6 +227,7 @@ int clp_enable_fh(struct zpci_dev *zdev, u32 *fh, u8 nr_dma_as);
  int clp_disable_fh(struct zpci_dev *zdev, u32 *fh);
  int clp_get_state(u32 fid, enum zpci_state *state);
  int clp_refresh_fh(u32 fid, u32 *fh);
+int zpci_get_mdd(u32 *mdd);
 
  /* UID */
  void update_uid_checking(bool new);
arch/s390/include/asm/pci_clp.h +1 -1
index 124fadfb74b9..d6bc324763f3 100644
@@ -76,7 +76,7 @@ struct clp_req_list_pci {
  struct clp_rsp_list_pci {
  	struct clp_rsp_hdr hdr;
  	u64 resume_token;
-	u32 reserved2;
+	u32 mdd;
  	u16 max_fn;
  	u8			: 7;
  	u8 uid_checking		: 1;
arch/s390/pci/pci_clp.c +5 -3
index bc7446566cbc..308ffb93413f 100644
@@ -328,7 +328,7 @@ int clp_disable_fh(struct zpci_dev *zdev, u32 *fh)
  }
 
  static int clp_list_pci_req(struct clp_req_rsp_list_pci *rrb,
-			    u64 *resume_token, int *nentries)
+			    u64 *resume_token, int *nentries, u32 *mdd)
  {
  	int rc;
 
@@ -354,6 +354,8 @@ static int clp_list_pci_req(struct clp_req_rsp_list_pci *rrb,
  	*nentries = (rrb->response.hdr.len - LIST_PCI_HDR_LEN) /
  		rrb->response.entry_size;
  	*resume_token = rrb->response.resume_token;
+	if (mdd)
+		*mdd = rrb->response.mdd;
 
  	return rc;
  }
@@ -365,7 +367,7 @@ static int clp_list_pci(struct clp_req_rsp_list_pci *rrb, void *data,
  	int nentries, i, rc;
 
  	do {
-		rc = clp_list_pci_req(rrb, &resume_token, &nentries);
+		rc = clp_list_pci_req(rrb, &resume_token, &nentries, NULL);
  		if (rc)
  			return rc;
  		for (i = 0; i < nentries; i++)
@@ -383,7 +385,7 @@ static int clp_find_pci(struct clp_req_rsp_list_pci *rrb, u32 fid,
  	int nentries, i, rc;
 
  	do {
-		rc = clp_list_pci_req(rrb, &resume_token, &nentries);
+		rc = clp_list_pci_req(rrb, &resume_token, &nentries, NULL);
  		if (rc)
  			return rc;
  		fh_list = rrb->response.fh_list;
@@ -468,6 +470,26 @@ int clp_get_state(u32 fid, enum zpci_state *state)
  	return rc;
  }
 
Quoted 12 lines

arch/s390/pci/pci_clp.c int clp_get_state(u32 fid, enum zpci_state *state) · line 480

Quoted 8 lines
arch/s390/pci/pci_clp.c +8 -0
+int zpci_get_mdd(u32 *mdd)
+{
+	struct clp_req_rsp_list_pci *rrb;
+	u64 resume_token = 0;
+	int nentries, rc;
+
+	if (!mdd)
+		return -EINVAL;  

I think this tests is not useful.
The caller must take care not to call with a NULL pointer, what the only caller today make sure.

what if the caller does it anyway?

I think the test is useful. if passing NULL is a bug, then maybe consider using BUG_ON, or WARN_ONCE

Quoted 19 lines
Quoted 18 lines
arch/s390/pci/pci_clp.c +12 -0
+
+	rrb = clp_alloc_block(GFP_KERNEL);
+	if (!rrb)
+		return -ENOMEM;
+
+	rc = clp_list_pci_req(rrb, &resume_token, &nentries, mdd);
+
+	clp_free_block(rrb);
+	return rc;
+}
+EXPORT_SYMBOL_GPL(zpci_get_mdd);
+
  static int clp_base_slpc(struct clp_req *req, struct clp_req_rsp_slpc *lpcb)
  {
  	unsigned long limit = PAGE_SIZE - sizeof(lpcb->request);
#27 lore raw

Re: [PATCH v2 12/30] s390/pci: get SHM information from list pci

Quoted 98 lines · Claudio Imbrenda
Quoted 82 lines · On Tue, 18 Jan 2022 11:36:06 +0100 Pierre Morel <[email protected]> wrote:
Quoted 81 lines · Matthew Rosato

KVM will need information on the special handle mask used to indicate emulated devices. In order to obtain this, a new type of list pci call must be made to gather the information. Extend clp_list_pci_req to also fetch the model-dependent-data field that holds this mask.

---
   arch/s390/include/asm/pci.h     |  1 +
   arch/s390/include/asm/pci_clp.h |  2 +-
   arch/s390/pci/pci_clp.c         | 28 +++++++++++++++++++++++++---
   3 files changed, 27 insertions(+), 4 deletions(-)
arch/s390/include/asm/pci.h +1 -0
index 00a2c24d6d2b..f3cd2da8128c 100644
@@ -227,6 +227,7 @@ int clp_enable_fh(struct zpci_dev *zdev, u32 *fh, u8 nr_dma_as);
   int clp_disable_fh(struct zpci_dev *zdev, u32 *fh);
   int clp_get_state(u32 fid, enum zpci_state *state);
   int clp_refresh_fh(u32 fid, u32 *fh);
+int zpci_get_mdd(u32 *mdd);
 
   /* UID */
   void update_uid_checking(bool new);
arch/s390/include/asm/pci_clp.h +1 -1
index 124fadfb74b9..d6bc324763f3 100644
@@ -76,7 +76,7 @@ struct clp_req_list_pci {
   struct clp_rsp_list_pci {
   	struct clp_rsp_hdr hdr;
   	u64 resume_token;
-	u32 reserved2;
+	u32 mdd;
   	u16 max_fn;
   	u8			: 7;
   	u8 uid_checking		: 1;
arch/s390/pci/pci_clp.c +5 -3
index bc7446566cbc..308ffb93413f 100644
@@ -328,7 +328,7 @@ int clp_disable_fh(struct zpci_dev *zdev, u32 *fh)
   }
 
   static int clp_list_pci_req(struct clp_req_rsp_list_pci *rrb,
-			    u64 *resume_token, int *nentries)
+			    u64 *resume_token, int *nentries, u32 *mdd)
   {
   	int rc;
 
@@ -354,6 +354,8 @@ static int clp_list_pci_req(struct clp_req_rsp_list_pci *rrb,
   	*nentries = (rrb->response.hdr.len - LIST_PCI_HDR_LEN) /
   		rrb->response.entry_size;
   	*resume_token = rrb->response.resume_token;
+	if (mdd)
+		*mdd = rrb->response.mdd;
 
   	return rc;
   }
@@ -365,7 +367,7 @@ static int clp_list_pci(struct clp_req_rsp_list_pci *rrb, void *data,
   	int nentries, i, rc;
 
   	do {
-		rc = clp_list_pci_req(rrb, &resume_token, &nentries);
+		rc = clp_list_pci_req(rrb, &resume_token, &nentries, NULL);
   		if (rc)
   			return rc;
   		for (i = 0; i < nentries; i++)
@@ -383,7 +385,7 @@ static int clp_find_pci(struct clp_req_rsp_list_pci *rrb, u32 fid,
   	int nentries, i, rc;
 
   	do {
-		rc = clp_list_pci_req(rrb, &resume_token, &nentries);
+		rc = clp_list_pci_req(rrb, &resume_token, &nentries, NULL);
   		if (rc)
   			return rc;
   		fh_list = rrb->response.fh_list;
@@ -468,6 +470,26 @@ int clp_get_state(u32 fid, enum zpci_state *state)
   	return rc;
   }
 
Quoted 12 lines

arch/s390/pci/pci_clp.c int clp_get_state(u32 fid, enum zpci_state *state) · line 480

Quoted 8 lines
arch/s390/pci/pci_clp.c +8 -0
+int zpci_get_mdd(u32 *mdd)
+{
+	struct clp_req_rsp_list_pci *rrb;
+	u64 resume_token = 0;
+	int nentries, rc;
+
+	if (!mdd)
+		return -EINVAL;

I think this tests is not useful.
The caller must take care not to call with a NULL pointer, what the only caller today make sure.

Quoted 4 lines

what if the caller does it anyway?

I think the test is useful. if passing NULL is a bug, then maybe consider using BUG_ON, or WARN_ONCE

I think generally the caller is responsible for the test.
In our case for example the caller can use directly the address
of a u32 allocated on the stack or globaly and he knows if a test is useful or not.

Of course we can systematically check in every kernel function all pointer parameters against NULL.
But this is not userland, not even a inter-architecture core function and we can expect the kernel programmer to programm correctly.

For our special case zpci_get_mdd() nor clp_list_pci_req() do access *mdd if mdd is NULL so that giving a NULL mdd pointer will not trigger a fault.
Also, the function is named zpci_get_mdd(u32 *mdd) if the caller do not give a pointer to mdd which would be quite stupid to call a function zpci_get_mdd() in this circumstance he will just no get mdd, no side effect. So the only purpose having this test here is to say the caller that he forgot to check his mdd allocation.
My opinion is that he should have check.

Quoted 21 lines
Quoted 20 lines
Quoted 19 lines
arch/s390/pci/pci_clp.c +12 -0
+
+	rrb = clp_alloc_block(GFP_KERNEL);
+	if (!rrb)
+		return -ENOMEM;
+
+	rc = clp_list_pci_req(rrb, &resume_token, &nentries, mdd);
+
+	clp_free_block(rrb);
+	return rc;
+}
+EXPORT_SYMBOL_GPL(zpci_get_mdd);
+
   static int clp_base_slpc(struct clp_req *req, struct clp_req_rsp_slpc *lpcb)
   {
   	unsigned long limit = PAGE_SIZE - sizeof(lpcb->request);
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#28 lore raw

Re: [PATCH v2 12/30] s390/pci: get SHM information from list pci

Quoted 103 lines · Pierre Morel
Quoted 95 lines · Claudio Imbrenda
Quoted 83 lines · On Tue, 18 Jan 2022 11:36:06 +0100 Pierre Morel <[email protected]> wrote:
Quoted 82 lines · Matthew Rosato

KVM will need information on the special handle mask used to indicate emulated devices.  In order to obtain this, a new type of list pci call must be made to gather the information.  Extend clp_list_pci_req to also fetch the model-dependent-data field that holds this mask.

---
   arch/s390/include/asm/pci.h     |  1 + arch/s390/include/asm/pci_clp.h |  2 +- arch/s390/pci/pci_clp.c         | 28 +++++++++++++++++++++++++---

   3 files changed, 27 insertions(+), 4 deletions(-)
arch/s390/include/asm/pci.h +1 -0
index 00a2c24d6d2b..f3cd2da8128c 100644
@@ -227,6 +227,7 @@ int clp_enable_fh(struct zpci_dev *zdev, u32 
 *fh, u8 nr_dma_as);
    int clp_disable_fh(struct zpci_dev *zdev, u32 *fh);
    int clp_get_state(u32 fid, enum zpci_state *state);
    int clp_refresh_fh(u32 fid, u32 *fh);
+int zpci_get_mdd(u32 *mdd);
    /* UID */
    void update_uid_checking(bool new);
arch/s390/include/asm/pci_clp.h +1 -1
index 124fadfb74b9..d6bc324763f3 100644
@@ -76,7 +76,7 @@ struct clp_req_list_pci {
    struct clp_rsp_list_pci {
        struct clp_rsp_hdr hdr;
        u64 resume_token;
-    u32 reserved2;
+    u32 mdd;
        u16 max_fn;
        u8            : 7;
        u8 uid_checking        : 1;
arch/s390/pci/pci_clp.c +9 -3
index bc7446566cbc..308ffb93413f 100644
@@ -328,7 +328,7 @@ int clp_disable_fh(struct zpci_dev *zdev, u32 *fh)
    }
    static int clp_list_pci_req(struct clp_req_rsp_list_pci *rrb,
-                u64 *resume_token, int *nentries)
+                u64 *resume_token, int *nentries, u32 *mdd)
    {
        int rc;
@@ -354,6 +354,8 @@ static int clp_list_pci_req(struct 
 clp_req_rsp_list_pci *rrb,
        *nentries = (rrb->response.hdr.len - LIST_PCI_HDR_LEN) /
            rrb->response.entry_size;
        *resume_token = rrb->response.resume_token;
+    if (mdd)
+        *mdd = rrb->response.mdd;
        return rc;
    }
@@ -365,7 +367,7 @@ static int clp_list_pci(struct 
 clp_req_rsp_list_pci *rrb, void *data,
        int nentries, i, rc;
        do {
-        rc = clp_list_pci_req(rrb, &resume_token, &nentries);
+        rc = clp_list_pci_req(rrb, &resume_token, &nentries, NULL);
            if (rc)
                return rc;
            for (i = 0; i < nentries; i++)
@@ -383,7 +385,7 @@ static int clp_find_pci(struct 
 clp_req_rsp_list_pci *rrb, u32 fid,
        int nentries, i, rc;
        do {
-        rc = clp_list_pci_req(rrb, &resume_token, &nentries);
+        rc = clp_list_pci_req(rrb, &resume_token, &nentries, NULL);
            if (rc)
                return rc;
            fh_list = rrb->response.fh_list;
@@ -468,6 +470,26 @@ int clp_get_state(u32 fid, enum zpci_state *state)
        return rc;
    }
+int zpci_get_mdd(u32 *mdd)
+{
+    struct clp_req_rsp_list_pci *rrb;
+    u64 resume_token = 0;
Quoted 8 lines

arch/s390/pci/pci_clp.c int clp_get_state(u32 fid, enum zpci_state *state) · line 480

Quoted 4 lines
arch/s390/pci/pci_clp.c +4 -0
+    int nentries, rc;
+
+    if (!mdd)
+        return -EINVAL;

I think this tests is not useful.
The caller must take care not to call with a NULL pointer, what the only caller today make sure.

Quoted 4 lines

what if the caller does it anyway?

I think the test is useful. if passing NULL is a bug, then maybe consider using BUG_ON, or WARN_ONCE

Quoted 6 lines

I think generally the caller is responsible for the test.
In our case for example the caller can use directly the address
of a u32 allocated on the stack or globaly and he knows if a test is useful or not.

Of course we can systematically check in every kernel function all pointer parameters against NULL.
But this is not userland, not even a inter-architecture core function and we can expect the kernel programmer to programm correctly.

I appreciate your optimism :)

Quoted 5 lines

For our special case zpci_get_mdd() nor clp_list_pci_req() do access *mdd if mdd is NULL so that giving a NULL mdd pointer will not trigger a fault.
Also, the function is named zpci_get_mdd(u32 *mdd) if the caller do not give a pointer to mdd which would be quite stupid to call a function zpci_get_mdd() in this circumstance he will just no get mdd, no side effect. So the only purpose having this test here is to say the caller that he forgot to check his mdd allocation.
My opinion is that he should have check.

Based on the thread of conversation, I'm going to assume you meant 'My opinion that he should -not- have the check here'

So, I'm generally a fan of being a bit defensive in paths that are not performance-intensive and would therefore typically tend to agree with Claudio. But in this particular case, I'm willing to just drop this check and move on, primarily because 1) we only have a single caller already checking this case and 2) I don't really anticipate any additional callers later.

@Niklas do you care / is it OK to keep your R-b if I remove this if(!mdd) check?

maintainer Reviewed-by on v2 12/30
#29 lore raw

Re: [PATCH v2 12/30] s390/pci: get SHM information from list pci

In reply to Matthew Rosato
Quoted 5 lines · Matthew Rosato

KVM will need information on the special handle mask used to indicate emulated devices. In order to obtain this, a new type of list pci call must be made to gather the information. Extend clp_list_pci_req to also fetch the model-dependent-data field that holds this mask.

Quoted 7 lines
---
 arch/s390/include/asm/pci.h     |  1 +
 arch/s390/include/asm/pci_clp.h |  2 +-
 arch/s390/pci/pci_clp.c         | 28 +++++++++++++++++++++++++---
 3 files changed, 27 insertions(+), 4 deletions(-)

---8<---

Quoted 14 lines
arch/s390/pci/pci_clp.c +13 -0
+int zpci_get_mdd(u32 *mdd)
+{
+	struct clp_req_rsp_list_pci *rrb;
+	u64 resume_token = 0;
+	int nentries, rc;
+
+	if (!mdd)
+		return -EINVAL;
+
+	rrb = clp_alloc_block(GFP_KERNEL);
+	if (!rrb)
+		return -ENOMEM;
+

arch/s390/pci/pci_clp.c int clp_get_state(u32 fid, enum zpci_state *state) · line 495

Quoted 10 lines
arch/s390/pci/pci_clp.c +7 -0
+	rc = clp_list_pci_req(rrb, &resume_token, &nentries, mdd);
+
+	clp_free_block(rrb);
+	return rc;
+}
+EXPORT_SYMBOL_GPL(zpci_get_mdd);
+
 static int clp_base_slpc(struct clp_req *req, struct clp_req_rsp_slpc *lpcb)
 {
 	unsigned long limit = PAGE_SIZE - sizeof(lpcb->request);

Looks good.

author maintainer
#30 lore raw

PATCH v2 13/30 s390/pci: return status from zpci_refresh_trans

In reply to Matthew Rosato

Current callers of zpci_refresh_trans don't need to interrogate the status returned from the underlying instructions. However, a subsequent patch will add a KVM caller that needs this information. Add a new argument to zpci_refresh_trans to pass the address of a status byte and update existing call sites to provide it.

---
 arch/s390/include/asm/pci_insn.h |  2 +-
 arch/s390/pci/pci_dma.c          |  6 ++++--
 arch/s390/pci/pci_insn.c         | 10 +++++-----
 drivers/iommu/s390-iommu.c       |  4 +++-
 4 files changed, 13 insertions(+), 9 deletions(-)
arch/s390/include/asm/pci_insn.h +1 -1
index 5331082fa516..32759c407b8f 100644
@@ -135,7 +135,7 @@ union zpci_sic_iib {
 DECLARE_STATIC_KEY_FALSE(have_mio);
 
 u8 zpci_mod_fc(u64 req, struct zpci_fib *fib, u8 *status);
-int zpci_refresh_trans(u64 fn, u64 addr, u64 range);
+int zpci_refresh_trans(u64 fn, u64 addr, u64 range, u8 *status);
 int __zpci_load(u64 *data, u64 req, u64 offset);
 int zpci_load(u64 *data, const volatile void __iomem *addr, unsigned long len);
 int __zpci_store(u64 data, u64 req, u64 offset);
arch/s390/pci/pci_dma.c +4 -2
index a81de48d5ea7..b0a2380bcad8 100644
@@ -23,8 +23,9 @@ static u32 s390_iommu_aperture_factor = 1;
 
 static int zpci_refresh_global(struct zpci_dev *zdev)
 {
+	u8 status;
 	return zpci_refresh_trans((u64) zdev->fh << 32, zdev->start_dma,
-				  zdev->iommu_pages * PAGE_SIZE);
+				  zdev->iommu_pages * PAGE_SIZE, &status);
 }
 
 unsigned long *dma_alloc_cpu_table(void)
@@ -183,6 +184,7 @@ static int __dma_purge_tlb(struct zpci_dev *zdev, dma_addr_t dma_addr,
 			   size_t size, int flags)
 {
 	unsigned long irqflags;
+	u8 status;
 	int ret;
 
 	/*
@@ -201,7 +203,7 @@ static int __dma_purge_tlb(struct zpci_dev *zdev, dma_addr_t dma_addr,
 	}
 
 	ret = zpci_refresh_trans((u64) zdev->fh << 32, dma_addr,
-				 PAGE_ALIGN(size));
+				 PAGE_ALIGN(size), &status);
 	if (ret == -ENOMEM && !s390_iommu_strict) {
 		/* enable the hypervisor to free some resources */
 		if (zpci_refresh_global(zdev))
arch/s390/pci/pci_insn.c +5 -5
index 0509554301c7..ca6399d52767 100644
@@ -77,20 +77,20 @@ static inline u8 __rpcit(u64 fn, u64 addr, u64 range, u8 *status)
 	return cc;
 }
 
-int zpci_refresh_trans(u64 fn, u64 addr, u64 range)
+int zpci_refresh_trans(u64 fn, u64 addr, u64 range, u8 *status)
 {
-	u8 cc, status;
+	u8 cc;
 
 	do {
-		cc = __rpcit(fn, addr, range, &status);
+		cc = __rpcit(fn, addr, range, status);
 		if (cc == 2)
 			udelay(ZPCI_INSN_BUSY_DELAY);
 	} while (cc == 2);
 
 	if (cc)
-		zpci_err_insn(cc, status, addr, range);
+		zpci_err_insn(cc, *status, addr, range);
 
-	if (cc == 1 && (status == 4 || status == 16))
+	if (cc == 1 && (*status == 4 || *status == 16))
 		return -ENOMEM;
 
 	return (cc) ? -EIO : 0;
drivers/iommu/s390-iommu.c +3 -1
index 50860ebdd087..845bb99c183e 100644
@@ -214,6 +214,7 @@ static int s390_iommu_update_trans(struct s390_domain *s390_domain,
 	unsigned long irq_flags, nr_pages, i;
 	unsigned long *entry;
 	int rc = 0;
+	u8 status;
 
 	if (dma_addr < s390_domain->domain.geometry.aperture_start ||
 	    dma_addr + size > s390_domain->domain.geometry.aperture_end)
@@ -238,7 +239,8 @@ static int s390_iommu_update_trans(struct s390_domain *s390_domain,
 	spin_lock(&s390_domain->list_lock);
 	list_for_each_entry(domain_device, &s390_domain->devices, list) {
 		rc = zpci_refresh_trans((u64) domain_device->zdev->fh << 32,
-					start_dma_addr, nr_pages * PAGE_SIZE);
+					start_dma_addr, nr_pages * PAGE_SIZE,
+					&status);
 		if (rc)
 			break;
 	}
Signature · 1 line
2.27.0
Reviewed-by on v2 13/30
#31 lore raw

Re: [PATCH v2 13/30] s390/pci: return status from zpci_refresh_trans

Quoted 7 lines · Matthew Rosato

Current callers of zpci_refresh_trans don't need to interrogate the status returned from the underlying instructions. However, a subsequent patch will add a KVM caller that needs this information. Add a new argument to zpci_refresh_trans to pass the address of a status byte and update existing call sites to provide it.

Quoted 105 lines
---
  arch/s390/include/asm/pci_insn.h |  2 +-
  arch/s390/pci/pci_dma.c          |  6 ++++--
  arch/s390/pci/pci_insn.c         | 10 +++++-----
  drivers/iommu/s390-iommu.c       |  4 +++-
  4 files changed, 13 insertions(+), 9 deletions(-)
arch/s390/include/asm/pci_insn.h +1 -1
index 5331082fa516..32759c407b8f 100644
@@ -135,7 +135,7 @@ union zpci_sic_iib {
  DECLARE_STATIC_KEY_FALSE(have_mio);
 
  u8 zpci_mod_fc(u64 req, struct zpci_fib *fib, u8 *status);
-int zpci_refresh_trans(u64 fn, u64 addr, u64 range);
+int zpci_refresh_trans(u64 fn, u64 addr, u64 range, u8 *status);
  int __zpci_load(u64 *data, u64 req, u64 offset);
  int zpci_load(u64 *data, const volatile void __iomem *addr, unsigned long len);
  int __zpci_store(u64 data, u64 req, u64 offset);
arch/s390/pci/pci_dma.c +4 -2
index a81de48d5ea7..b0a2380bcad8 100644
@@ -23,8 +23,9 @@ static u32 s390_iommu_aperture_factor = 1;
 
  static int zpci_refresh_global(struct zpci_dev *zdev)
  {
+	u8 status;
  	return zpci_refresh_trans((u64) zdev->fh << 32, zdev->start_dma,
-				  zdev->iommu_pages * PAGE_SIZE);
+				  zdev->iommu_pages * PAGE_SIZE, &status);
  }
 
  unsigned long *dma_alloc_cpu_table(void)
@@ -183,6 +184,7 @@ static int __dma_purge_tlb(struct zpci_dev *zdev, dma_addr_t dma_addr,
  			   size_t size, int flags)
  {
  	unsigned long irqflags;
+	u8 status;
  	int ret;
 
  	/*
@@ -201,7 +203,7 @@ static int __dma_purge_tlb(struct zpci_dev *zdev, dma_addr_t dma_addr,
  	}
 
  	ret = zpci_refresh_trans((u64) zdev->fh << 32, dma_addr,
-				 PAGE_ALIGN(size));
+				 PAGE_ALIGN(size), &status);
  	if (ret == -ENOMEM && !s390_iommu_strict) {
  		/* enable the hypervisor to free some resources */
  		if (zpci_refresh_global(zdev))
arch/s390/pci/pci_insn.c +5 -5
index 0509554301c7..ca6399d52767 100644
@@ -77,20 +77,20 @@ static inline u8 __rpcit(u64 fn, u64 addr, u64 range, u8 *status)
  	return cc;
  }
 
-int zpci_refresh_trans(u64 fn, u64 addr, u64 range)
+int zpci_refresh_trans(u64 fn, u64 addr, u64 range, u8 *status)
  {
-	u8 cc, status;
+	u8 cc;
 
  	do {
-		cc = __rpcit(fn, addr, range, &status);
+		cc = __rpcit(fn, addr, range, status);
  		if (cc == 2)
  			udelay(ZPCI_INSN_BUSY_DELAY);
  	} while (cc == 2);
 
  	if (cc)
-		zpci_err_insn(cc, status, addr, range);
+		zpci_err_insn(cc, *status, addr, range);
 
-	if (cc == 1 && (status == 4 || status == 16))
+	if (cc == 1 && (*status == 4 || *status == 16))
  		return -ENOMEM;
 
  	return (cc) ? -EIO : 0;
drivers/iommu/s390-iommu.c +3 -1
index 50860ebdd087..845bb99c183e 100644
@@ -214,6 +214,7 @@ static int s390_iommu_update_trans(struct s390_domain *s390_domain,
  	unsigned long irq_flags, nr_pages, i;
  	unsigned long *entry;
  	int rc = 0;
+	u8 status;
 
  	if (dma_addr < s390_domain->domain.geometry.aperture_start ||
  	    dma_addr + size > s390_domain->domain.geometry.aperture_end)
@@ -238,7 +239,8 @@ static int s390_iommu_update_trans(struct s390_domain *s390_domain,
  	spin_lock(&s390_domain->list_lock);
  	list_for_each_entry(domain_device, &s390_domain->devices, list) {
  		rc = zpci_refresh_trans((u64) domain_device->zdev->fh << 32,
-					start_dma_addr, nr_pages * PAGE_SIZE);
+					start_dma_addr, nr_pages * PAGE_SIZE,
+					&status);
  		if (rc)
  			break;
  	}
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
maintainer Reviewed-by on v2 13/30
#32 lore raw

Re: [PATCH v2 13/30] s390/pci: return status from zpci_refresh_trans

In reply to Matthew Rosato
Quoted 7 lines · On Fri, 14 Jan 2022 15:31:28 -0500 Matthew Rosato <[email protected]>…

Current callers of zpci_refresh_trans don't need to interrogate the status returned from the underlying instructions. However, a subsequent patch will add a KVM caller that needs this information. Add a new argument to zpci_refresh_trans to pass the address of a status byte and update existing call sites to provide it.

Quoted 104 lines
---
 arch/s390/include/asm/pci_insn.h |  2 +-
 arch/s390/pci/pci_dma.c          |  6 ++++--
 arch/s390/pci/pci_insn.c         | 10 +++++-----
 drivers/iommu/s390-iommu.c       |  4 +++-
 4 files changed, 13 insertions(+), 9 deletions(-)
arch/s390/include/asm/pci_insn.h +1 -1
index 5331082fa516..32759c407b8f 100644
@@ -135,7 +135,7 @@ union zpci_sic_iib {
 DECLARE_STATIC_KEY_FALSE(have_mio);
 
 u8 zpci_mod_fc(u64 req, struct zpci_fib *fib, u8 *status);
-int zpci_refresh_trans(u64 fn, u64 addr, u64 range);
+int zpci_refresh_trans(u64 fn, u64 addr, u64 range, u8 *status);
 int __zpci_load(u64 *data, u64 req, u64 offset);
 int zpci_load(u64 *data, const volatile void __iomem *addr, unsigned long len);
 int __zpci_store(u64 data, u64 req, u64 offset);
arch/s390/pci/pci_dma.c +4 -2
index a81de48d5ea7..b0a2380bcad8 100644
@@ -23,8 +23,9 @@ static u32 s390_iommu_aperture_factor = 1;
 
 static int zpci_refresh_global(struct zpci_dev *zdev)
 {
+	u8 status;
 	return zpci_refresh_trans((u64) zdev->fh << 32, zdev->start_dma,
-				  zdev->iommu_pages * PAGE_SIZE);
+				  zdev->iommu_pages * PAGE_SIZE, &status);
 }
 
 unsigned long *dma_alloc_cpu_table(void)
@@ -183,6 +184,7 @@ static int __dma_purge_tlb(struct zpci_dev *zdev, dma_addr_t dma_addr,
 			   size_t size, int flags)
 {
 	unsigned long irqflags;
+	u8 status;
 	int ret;
 
 	/*
@@ -201,7 +203,7 @@ static int __dma_purge_tlb(struct zpci_dev *zdev, dma_addr_t dma_addr,
 	}
 
 	ret = zpci_refresh_trans((u64) zdev->fh << 32, dma_addr,
-				 PAGE_ALIGN(size));
+				 PAGE_ALIGN(size), &status);
 	if (ret == -ENOMEM && !s390_iommu_strict) {
 		/* enable the hypervisor to free some resources */
 		if (zpci_refresh_global(zdev))
arch/s390/pci/pci_insn.c +5 -5
index 0509554301c7..ca6399d52767 100644
@@ -77,20 +77,20 @@ static inline u8 __rpcit(u64 fn, u64 addr, u64 range, u8 *status)
 	return cc;
 }
 
-int zpci_refresh_trans(u64 fn, u64 addr, u64 range)
+int zpci_refresh_trans(u64 fn, u64 addr, u64 range, u8 *status)
 {
-	u8 cc, status;
+	u8 cc;
 
 	do {
-		cc = __rpcit(fn, addr, range, &status);
+		cc = __rpcit(fn, addr, range, status);
 		if (cc == 2)
 			udelay(ZPCI_INSN_BUSY_DELAY);
 	} while (cc == 2);
 
 	if (cc)
-		zpci_err_insn(cc, status, addr, range);
+		zpci_err_insn(cc, *status, addr, range);
 
-	if (cc == 1 && (status == 4 || status == 16))
+	if (cc == 1 && (*status == 4 || *status == 16))
 		return -ENOMEM;
 
 	return (cc) ? -EIO : 0;
drivers/iommu/s390-iommu.c +3 -1
index 50860ebdd087..845bb99c183e 100644
@@ -214,6 +214,7 @@ static int s390_iommu_update_trans(struct s390_domain *s390_domain,
 	unsigned long irq_flags, nr_pages, i;
 	unsigned long *entry;
 	int rc = 0;
+	u8 status;
 
 	if (dma_addr < s390_domain->domain.geometry.aperture_start ||
 	    dma_addr + size > s390_domain->domain.geometry.aperture_end)
@@ -238,7 +239,8 @@ static int s390_iommu_update_trans(struct s390_domain *s390_domain,
 	spin_lock(&s390_domain->list_lock);
 	list_for_each_entry(domain_device, &s390_domain->devices, list) {
 		rc = zpci_refresh_trans((u64) domain_device->zdev->fh << 32,
-					start_dma_addr, nr_pages * PAGE_SIZE);
+					start_dma_addr, nr_pages * PAGE_SIZE,
+					&status);
 		if (rc)
 			break;
 	}
maintainer Reviewed-by on v2 13/30
#33 lore raw

Re: [PATCH v2 13/30] s390/pci: return status from zpci_refresh_trans

In reply to Matthew Rosato
Quoted 19 lines · Matthew Rosato

Current callers of zpci_refresh_trans don't need to interrogate the status returned from the underlying instructions. However, a subsequent patch will add a KVM caller that needs this information. Add a new argument to zpci_refresh_trans to pass the address of a status byte and update existing call sites to provide it.

---
 arch/s390/include/asm/pci_insn.h |  2 +-
 arch/s390/pci/pci_dma.c          |  6 ++++--
 arch/s390/pci/pci_insn.c         | 10 +++++-----
 drivers/iommu/s390-iommu.c       |  4 +++-
 4 files changed, 13 insertions(+), 9 deletions(-)
arch/s390/include/asm/pci_insn.h +0 -0
index 5331082fa516..32759c407b8f 100644

arch/s390/include/asm/pci_insn.h union zpci_sic_iib { · line 139

Quoted 6 lines
arch/s390/include/asm/pci_insn.h +1 -1
@@ -135,7 +135,7 @@ union zpci_sic_iib {
 DECLARE_STATIC_KEY_FALSE(have_mio);
 
 u8 zpci_mod_fc(u64 req, struct zpci_fib *fib, u8 *status);
-int zpci_refresh_trans(u64 fn, u64 addr, u64 range);
+int zpci_refresh_trans(u64 fn, u64 addr, u64 range, u8 *status);
 

---8<---

Quoted 15 lines
arch/s390/include/asm/pci_insn.h +0 -0
 	return (cc) ? -EIO : 0;
drivers/iommu/s390-iommu.c +1 -0
index 50860ebdd087..845bb99c183e 100644
@@ -214,6 +214,7 @@ static int s390_iommu_update_trans(struct s390_domain *s390_domain,
 	unsigned long irq_flags, nr_pages, i;
 	unsigned long *entry;
 	int rc = 0;
+	u8 status;
 
 	if (dma_addr < s390_domain->domain.geometry.aperture_start ||
 	    dma_addr + size > s390_domain->domain.geometry.aperture_end)

drivers/iommu/s390-iommu.c static int s390_iommu_update_trans(struct s390_domain *s390_domain, · line 246

Quoted 9 lines
drivers/iommu/s390-iommu.c +2 -1
@@ -238,7 +239,8 @@ static int s390_iommu_update_trans(struct s390_domain *s390_domain,
 	spin_lock(&s390_domain->list_lock);
 	list_for_each_entry(domain_device, &s390_domain->devices, list) {
 		rc = zpci_refresh_trans((u64) domain_device->zdev->fh << 32,
-					start_dma_addr, nr_pages * PAGE_SIZE);
+					start_dma_addr, nr_pages * PAGE_SIZE,
+					&status);
 		if (rc)
 			break;
 	}

Looks good.

author maintainer
#34 lore raw

PATCH v2 14/30 KVM: s390: pci: add basic kvm_zdev structure

In reply to Matthew Rosato

This structure will be used to carry kvm passthrough information related to zPCI devices.

---
 arch/s390/include/asm/kvm_pci.h | 29 +++++++++++++++++++++
 arch/s390/include/asm/pci.h     |  3 +++
 arch/s390/kvm/Makefile          |  2 +-
 arch/s390/kvm/pci.c             | 46 +++++++++++++++++++++++++++++++++
 4 files changed, 79 insertions(+), 1 deletion(-)
 create mode 100644 arch/s390/include/asm/kvm_pci.h
 create mode 100644 arch/s390/kvm/pci.c
arch/s390/include/asm/kvm_pci.h new file +29 -0
new file mode 100644
index 000000000000..aafee2976929
@@ -0,0 +1,29 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * KVM PCI Passthrough for virtual machines on s390
+ *
+ * Copyright IBM Corp. 2021
+ *
+ *    Author(s): Matthew Rosato <[email protected]>
+ */
+
+
+#ifndef ASM_KVM_PCI_H
+#define ASM_KVM_PCI_H
+
+#include <linux/types.h>
+#include <linux/kvm_types.h>
+#include <linux/kvm_host.h>
+#include <linux/kvm.h>
+#include <linux/pci.h>
+
+struct kvm_zdev {
+	struct zpci_dev *zdev;
+	struct kvm *kvm;
+};
+
+int kvm_s390_pci_dev_open(struct zpci_dev *zdev);
+void kvm_s390_pci_dev_release(struct zpci_dev *zdev);
+void kvm_s390_pci_attach_kvm(struct zpci_dev *zdev, struct kvm *kvm);
+
+#endif /* ASM_KVM_PCI_H */
arch/s390/include/asm/pci.h +3 -0
index f3cd2da8128c..9b6c657d8d31 100644
@@ -97,6 +97,7 @@ struct zpci_bar_struct {
 };
 
 struct s390_domain;
+struct kvm_zdev;
 
 #define ZPCI_FUNCTIONS_PER_BUS 256
 struct zpci_bus {
@@ -190,6 +191,8 @@ struct zpci_dev {
 	struct dentry	*debugfs_dev;
 
 	struct s390_domain *s390_domain; /* s390 IOMMU domain data */
+
+	struct kvm_zdev *kzdev; /* passthrough data */
 };
 
 static inline bool zdev_enabled(struct zpci_dev *zdev)
arch/s390/kvm/Makefile +1 -1
index b3aaadc60ead..a26f4fe7b680 100644
@@ -11,5 +11,5 @@ ccflags-y := -Ivirt/kvm -Iarch/s390/kvm
 
 kvm-objs := $(common-objs) kvm-s390.o intercept.o interrupt.o priv.o sigp.o
 kvm-objs += diag.o gaccess.o guestdbg.o vsie.o pv.o
-
+kvm-$(CONFIG_PCI) += pci.o
 obj-$(CONFIG_KVM) += kvm.o
arch/s390/kvm/pci.c new file +46 -0

46 lines not on this page · raw message

Signature · 1 line
2.27.0
Reviewed-by on v2 14/30
#35 lore raw

Re: [PATCH v2 14/30] KVM: s390: pci: add basic kvm_zdev structure

Quoted 19 lines · Matthew Rosato

This structure will be used to carry kvm passthrough information related to zPCI devices.

---
  arch/s390/include/asm/kvm_pci.h | 29 +++++++++++++++++++++
  arch/s390/include/asm/pci.h     |  3 +++
  arch/s390/kvm/Makefile          |  2 +-
  arch/s390/kvm/pci.c             | 46 +++++++++++++++++++++++++++++++++
  4 files changed, 79 insertions(+), 1 deletion(-)

create mode 100644 arch/s390/include/asm/kvm_pci.h create mode 100644 arch/s390/kvm/pci.c

arch/s390/include/asm/kvm_pci.h new file +0 -0
new file mode 100644
index 000000000000..aafee2976929

arch/s390/include/asm/kvm_pci.h line 10

Quoted 10 lines
arch/s390/include/asm/kvm_pci.h new file +10 -0
@@ -0,0 +1,29 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * KVM PCI Passthrough for virtual machines on s390
+ *
+ * Copyright IBM Corp. 2021
+ *
+ *    Author(s): Matthew Rosato <[email protected]>
+ */
+
+

One blank line too much.

Otherwise, look good to me.

Quoted 104 lines
arch/s390/include/asm/kvm_pci.h new file +19 -0
+#ifndef ASM_KVM_PCI_H
+#define ASM_KVM_PCI_H
+
+#include <linux/types.h>
+#include <linux/kvm_types.h>
+#include <linux/kvm_host.h>
+#include <linux/kvm.h>
+#include <linux/pci.h>
+
+struct kvm_zdev {
+	struct zpci_dev *zdev;
+	struct kvm *kvm;
+};
+
+int kvm_s390_pci_dev_open(struct zpci_dev *zdev);
+void kvm_s390_pci_dev_release(struct zpci_dev *zdev);
+void kvm_s390_pci_attach_kvm(struct zpci_dev *zdev, struct kvm *kvm);
+
+#endif /* ASM_KVM_PCI_H */
arch/s390/include/asm/pci.h +3 -0
index f3cd2da8128c..9b6c657d8d31 100644
@@ -97,6 +97,7 @@ struct zpci_bar_struct {
  };
 
  struct s390_domain;
+struct kvm_zdev;
 
  #define ZPCI_FUNCTIONS_PER_BUS 256
  struct zpci_bus {
@@ -190,6 +191,8 @@ struct zpci_dev {
  	struct dentry	*debugfs_dev;
 
  	struct s390_domain *s390_domain; /* s390 IOMMU domain data */
+
+	struct kvm_zdev *kzdev; /* passthrough data */
  };
 
  static inline bool zdev_enabled(struct zpci_dev *zdev)
arch/s390/kvm/Makefile +1 -1
index b3aaadc60ead..a26f4fe7b680 100644
@@ -11,5 +11,5 @@ ccflags-y := -Ivirt/kvm -Iarch/s390/kvm
 
  kvm-objs := $(common-objs) kvm-s390.o intercept.o interrupt.o priv.o sigp.o
  kvm-objs += diag.o gaccess.o guestdbg.o vsie.o pv.o
-
+kvm-$(CONFIG_PCI) += pci.o
  obj-$(CONFIG_KVM) += kvm.o
arch/s390/kvm/pci.c new file +46 -0
new file mode 100644
index 000000000000..1c33bc7bf2bd
@@ -0,0 +1,46 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * s390 kvm PCI passthrough support
+ *
+ * Copyright IBM Corp. 2021
+ *
+ *    Author(s): Matthew Rosato <[email protected]>
+ */
+
+#include <linux/kvm_host.h>
+#include <linux/pci.h>
+#include <asm/kvm_pci.h>
+
+int kvm_s390_pci_dev_open(struct zpci_dev *zdev)
+{
+	struct kvm_zdev *kzdev;
+
+	kzdev = kzalloc(sizeof(struct kvm_zdev), GFP_KERNEL);
+	if (!kzdev)
+		return -ENOMEM;
+
+	kzdev->zdev = zdev;
+	zdev->kzdev = kzdev;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(kvm_s390_pci_dev_open);
+
+void kvm_s390_pci_dev_release(struct zpci_dev *zdev)
+{
+	struct kvm_zdev *kzdev;
+
+	kzdev = zdev->kzdev;
+	WARN_ON(kzdev->zdev != zdev);
+	zdev->kzdev = 0;
+	kfree(kzdev);
+}
+EXPORT_SYMBOL_GPL(kvm_s390_pci_dev_release);
+
+void kvm_s390_pci_attach_kvm(struct zpci_dev *zdev, struct kvm *kvm)
+{
+	struct kvm_zdev *kzdev = zdev->kzdev;
+
+	kzdev->kvm = kvm;
+}
+EXPORT_SYMBOL_GPL(kvm_s390_pci_attach_kvm);
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
#36 lore raw

Re: [PATCH v2 14/30] KVM: s390: pci: add basic kvm_zdev structure

Quoted 130 lines · Pierre Morel
Quoted 19 lines · Matthew Rosato

This structure will be used to carry kvm passthrough information related to zPCI devices.

---
  arch/s390/include/asm/kvm_pci.h | 29 +++++++++++++++++++++ arch/s390/include/asm/pci.h     |  3 +++ arch/s390/kvm/Makefile          |  2 +- arch/s390/kvm/pci.c             | 46 +++++++++++++++++++++++++++++++++

  4 files changed, 79 insertions(+), 1 deletion(-)

  create mode 100644 arch/s390/include/asm/kvm_pci.h create mode 100644 arch/s390/kvm/pci.c

arch/s390/include/asm/kvm_pci.h new file +0 -0
new file mode 100644
index 000000000000..aafee2976929

arch/s390/include/asm/kvm_pci.h line 10

Quoted 10 lines
arch/s390/include/asm/kvm_pci.h new file +10 -0
@@ -0,0 +1,29 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * KVM PCI Passthrough for virtual machines on s390
+ *
+ * Copyright IBM Corp. 2021
+ *
+ *    Author(s): Matthew Rosato <[email protected]>
+ */
+
+

One blank line too much.

Otherwise, look good to me.

Quoted 91 lines
arch/s390/include/asm/kvm_pci.h new file +19 -0
+#ifndef ASM_KVM_PCI_H
+#define ASM_KVM_PCI_H
+
+#include <linux/types.h>
+#include <linux/kvm_types.h>
+#include <linux/kvm_host.h>
+#include <linux/kvm.h>
+#include <linux/pci.h>
+
+struct kvm_zdev {
+    struct zpci_dev *zdev;
+    struct kvm *kvm;
+};
+
+int kvm_s390_pci_dev_open(struct zpci_dev *zdev);
+void kvm_s390_pci_dev_release(struct zpci_dev *zdev);
+void kvm_s390_pci_attach_kvm(struct zpci_dev *zdev, struct kvm *kvm);
+
+#endif /* ASM_KVM_PCI_H */
arch/s390/include/asm/pci.h +1 -0
index f3cd2da8128c..9b6c657d8d31 100644
@@ -97,6 +97,7 @@ struct zpci_bar_struct {
   };
   struct s390_domain;
+struct kvm_zdev;
   #define ZPCI_FUNCTIONS_PER_BUS 256
   struct zpci_bus {
@@ -190,6 +191,8 @@ struct zpci_dev {
       struct dentry    *debugfs_dev;

      struct s390_domain *s390_domain; /* s390 IOMMU domain data */ +
+    struct kvm_zdev *kzdev; /* passthrough data */
  };
  static inline bool zdev_enabled(struct zpci_dev *zdev)

arch/s390/kvm/Makefile +0 -0
index b3aaadc60ead..a26f4fe7b680 100644
@@ -11,5 +11,5 @@ ccflags-y := -Ivirt/kvm -Iarch/s390/kvm

  kvm-objs := $(common-objs) kvm-s390.o intercept.o interrupt.o priv.o sigp.o kvm-objs += diag.o gaccess.o guestdbg.o vsie.o pv.o

arch/s390/kvm/Makefile +1 -1
-
+kvm-$(CONFIG_PCI) += pci.o

  obj-$(CONFIG_KVM) += kvm.o

arch/s390/kvm/pci.c new file +39 -0
new file mode 100644
index 000000000000..1c33bc7bf2bd
@@ -0,0 +1,46 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * s390 kvm PCI passthrough support
+ *
+ * Copyright IBM Corp. 2021
+ *
+ *    Author(s): Matthew Rosato <[email protected]>
+ */
+
+#include <linux/kvm_host.h>
+#include <linux/pci.h>
+#include <asm/kvm_pci.h>
+
+int kvm_s390_pci_dev_open(struct zpci_dev *zdev)
+{
+    struct kvm_zdev *kzdev;
+
+    kzdev = kzalloc(sizeof(struct kvm_zdev), GFP_KERNEL);
+    if (!kzdev)
+        return -ENOMEM;
+
+    kzdev->zdev = zdev;
+    zdev->kzdev = kzdev;
+
+    return 0;
+}
+EXPORT_SYMBOL_GPL(kvm_s390_pci_dev_open);
+
+void kvm_s390_pci_dev_release(struct zpci_dev *zdev)
+{
+    struct kvm_zdev *kzdev;
+
+    kzdev = zdev->kzdev;
+    WARN_ON(kzdev->zdev != zdev);
+    zdev->kzdev = 0;
+    kfree(kzdev);
+}
+EXPORT_SYMBOL_GPL(kvm_s390_pci_dev_release);
+
Quoted 9 lines

arch/s390/kvm/pci.c line 46

Quoted 8 lines
arch/s390/kvm/pci.c new file +7 -0
+void kvm_s390_pci_attach_kvm(struct zpci_dev *zdev, struct kvm *kvm)
+{
+    struct kvm_zdev *kzdev = zdev->kzdev;
+
+    kzdev->kvm = kvm;
+}
+EXPORT_SYMBOL_GPL(kvm_s390_pci_attach_kvm);

Working now on patch 24, I am not sure that this function is necessary.
the only purpose seems to set kzdev->kvm = kvm while we already know kzdev in the caller.

Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#37 lore raw

Re: [PATCH v2 14/30] KVM: s390: pci: add basic kvm_zdev structure

Quoted 79 lines · Pierre Morel
Quoted 76 lines · Pierre Morel
Quoted 19 lines · Matthew Rosato

This structure will be used to carry kvm passthrough information related to zPCI devices.

---
  arch/s390/include/asm/kvm_pci.h | 29 +++++++++++++++++++++ arch/s390/include/asm/pci.h     |  3 +++ arch/s390/kvm/Makefile          |  2 +- arch/s390/kvm/pci.c             | 46 +++++++++++++++++++++++++++++++++

  4 files changed, 79 insertions(+), 1 deletion(-)

  create mode 100644 arch/s390/include/asm/kvm_pci.h create mode 100644 arch/s390/kvm/pci.c

arch/s390/include/asm/kvm_pci.h new file +0 -0
new file mode 100644
index 000000000000..aafee2976929

arch/s390/include/asm/kvm_pci.h line 10

Quoted 10 lines
arch/s390/include/asm/kvm_pci.h new file +10 -0
@@ -0,0 +1,29 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * KVM PCI Passthrough for virtual machines on s390
+ *
+ * Copyright IBM Corp. 2021
+ *
+ *    Author(s): Matthew Rosato <[email protected]>
+ */
+
+

One blank line too much.

Otherwise, look good to me.

Quoted 37 lines
arch/s390/include/asm/kvm_pci.h new file +19 -0
+#ifndef ASM_KVM_PCI_H
+#define ASM_KVM_PCI_H
+
+#include <linux/types.h>
+#include <linux/kvm_types.h>
+#include <linux/kvm_host.h>
+#include <linux/kvm.h>
+#include <linux/pci.h>
+
+struct kvm_zdev {
+    struct zpci_dev *zdev;
+    struct kvm *kvm;
+};
+
+int kvm_s390_pci_dev_open(struct zpci_dev *zdev);
+void kvm_s390_pci_dev_release(struct zpci_dev *zdev);
+void kvm_s390_pci_attach_kvm(struct zpci_dev *zdev, struct kvm *kvm);
+
+#endif /* ASM_KVM_PCI_H */
arch/s390/include/asm/pci.h +1 -0
index f3cd2da8128c..9b6c657d8d31 100644
@@ -97,6 +97,7 @@ struct zpci_bar_struct {
   };
   struct s390_domain;
+struct kvm_zdev;
   #define ZPCI_FUNCTIONS_PER_BUS 256
   struct zpci_bus {
@@ -190,6 +191,8 @@ struct zpci_dev {
       struct dentry    *debugfs_dev;

      struct s390_domain *s390_domain; /* s390 IOMMU domain data */ +
+    struct kvm_zdev *kzdev; /* passthrough data */
  };
  static inline bool zdev_enabled(struct zpci_dev *zdev)

arch/s390/kvm/Makefile +0 -0
Quoted 8 lines
Quoted 8 lines

arch/s390/kvm/Makefile ccflags-y := -Ivirt/kvm -Iarch/s390/kvm · line 14

Quoted 8 lines

index b3aaadc60ead..a26f4fe7b680 100644

arch/s390/kvm/Makefile +0 -0
@@ -11,5 +11,5 @@ ccflags-y := -Ivirt/kvm -Iarch/s390/kvm

  kvm-objs := $(common-objs) kvm-s390.o intercept.o interrupt.o priv.o sigp.o kvm-objs += diag.o gaccess.o guestdbg.o vsie.o pv.o

arch/s390/kvm/Makefile +1 -1
-
+kvm-$(CONFIG_PCI) += pci.o

As discussed in other threads, I will look at changing this to kvm-$(CONFIG_VFIO_PCI_ZDEV) += pci.o
Along with other IS_ENABLE(CONFIG_PCI) -> IS_ENABLED(CONFIG_VFIO_PCI_ZDDEV) changes

Quoted 56 lines
Quoted 46 lines
Quoted 46 lines

  obj-$(CONFIG_KVM) += kvm.o

arch/s390/kvm/pci.c new file +39 -0
new file mode 100644
index 000000000000..1c33bc7bf2bd
@@ -0,0 +1,46 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * s390 kvm PCI passthrough support
+ *
+ * Copyright IBM Corp. 2021
+ *
+ *    Author(s): Matthew Rosato <[email protected]>
+ */
+
+#include <linux/kvm_host.h>
+#include <linux/pci.h>
+#include <asm/kvm_pci.h>
+
+int kvm_s390_pci_dev_open(struct zpci_dev *zdev)
+{
+    struct kvm_zdev *kzdev;
+
+    kzdev = kzalloc(sizeof(struct kvm_zdev), GFP_KERNEL);
+    if (!kzdev)
+        return -ENOMEM;
+
+    kzdev->zdev = zdev;
+    zdev->kzdev = kzdev;
+
+    return 0;
+}
+EXPORT_SYMBOL_GPL(kvm_s390_pci_dev_open);
+
+void kvm_s390_pci_dev_release(struct zpci_dev *zdev)
+{
+    struct kvm_zdev *kzdev;
+
+    kzdev = zdev->kzdev;
+    WARN_ON(kzdev->zdev != zdev);
+    zdev->kzdev = 0;
+    kfree(kzdev);
+}
+EXPORT_SYMBOL_GPL(kvm_s390_pci_dev_release);
+
Quoted 9 lines

arch/s390/kvm/pci.c line 46

Quoted 8 lines
arch/s390/kvm/pci.c new file +7 -0
+void kvm_s390_pci_attach_kvm(struct zpci_dev *zdev, struct kvm *kvm)
+{
+    struct kvm_zdev *kzdev = zdev->kzdev;
+
+    kzdev->kvm = kvm;
+}
+EXPORT_SYMBOL_GPL(kvm_s390_pci_attach_kvm);
Quoted 3 lines

Working now on patch 24, I am not sure that this function is necessary.
the only purpose seems to set kzdev->kvm = kvm while we already know kzdev in the caller.

Yep, as mentioned in the patch 24 thread I will drop this function and set kzdev->kvm = kvm directly in patch 24.

author maintainer
#38 lore raw

PATCH v2 15/30 KVM: s390: pci: do initial setup for AEN interpretation

In reply to Matthew Rosato

Initial setup for Adapter Event Notification Interpretation for zPCI passthrough devices. Specifically, allocate a structure for forwarding of adapter events and pass the address of this structure to firmware.

---
 arch/s390/include/asm/pci.h      |   4 +
 arch/s390/include/asm/pci_insn.h |  12 +++
 arch/s390/kvm/interrupt.c        |  14 +++
 arch/s390/kvm/kvm-s390.c         |   9 ++
 arch/s390/kvm/pci.c              | 144 +++++++++++++++++++++++++++++++
 arch/s390/kvm/pci.h              |  42 +++++++++
 arch/s390/pci/pci.c              |   6 ++
 7 files changed, 231 insertions(+)
 create mode 100644 arch/s390/kvm/pci.h
arch/s390/include/asm/pci.h +4 -0
index 9b6c657d8d31..9ff8dc19975e 100644
@@ -9,6 +9,7 @@
 #include <asm-generic/pci.h>
 #include <asm/pci_clp.h>
 #include <asm/pci_debug.h>
+#include <asm/pci_insn.h>
 #include <asm/sclp.h>
 
 #define PCIBIOS_MIN_IO		0x1000
@@ -204,6 +205,9 @@ extern const struct attribute_group *zpci_attr_groups[];
 extern unsigned int s390_pci_force_floating __initdata;
 extern unsigned int s390_pci_no_rid;
 
+extern union zpci_sic_iib *zpci_aipb;
+extern struct airq_iv *zpci_aif_sbv;
+
 /* -----------------------------------------------------------------------------
   Prototypes
 ----------------------------------------------------------------------------- */
arch/s390/include/asm/pci_insn.h +12 -0
index 32759c407b8f..ad9000295c82 100644
@@ -101,6 +101,7 @@ struct zpci_fib {
 /* Set Interruption Controls Operation Controls  */
 #define	SIC_IRQ_MODE_ALL		0
 #define	SIC_IRQ_MODE_SINGLE		1
+#define	SIC_SET_AENI_CONTROLS		2
 #define	SIC_IRQ_MODE_DIRECT		4
 #define	SIC_IRQ_MODE_D_ALL		16
 #define	SIC_IRQ_MODE_D_SINGLE		17
@@ -127,9 +128,20 @@ struct zpci_cdiib {
 	u64 : 64;
 } __packed __aligned(8);
 
+/* adapter interruption parameters block */
+struct zpci_aipb {
+	u64 faisb;
+	u64 gait;
+	u16 : 13;
+	u16 afi : 3;
+	u32 : 32;
+	u16 faal;
+} __packed __aligned(8);
+
 union zpci_sic_iib {
 	struct zpci_diib diib;
 	struct zpci_cdiib cdiib;
+	struct zpci_aipb aipb;
 };
 
 DECLARE_STATIC_KEY_FALSE(have_mio);
arch/s390/kvm/interrupt.c +14 -0
index f9b872e358c6..a591b8cd662f 100644
@@ -32,6 +32,7 @@
 #include "kvm-s390.h"
 #include "gaccess.h"
 #include "trace-s390.h"
+#include "pci.h"
 
 #define PFAULT_INIT 0x0600
 #define PFAULT_DONE 0x0680
@@ -3278,6 +3279,11 @@ void kvm_s390_gib_destroy(void)
 {
 	if (!gib)
 		return;
+	if (IS_ENABLED(CONFIG_PCI) && sclp.has_aeni && aift) {
+		mutex_lock(&aift->lock);
+		kvm_s390_pci_aen_exit();
+		mutex_unlock(&aift->lock);
+	}
 	chsc_sgib(0);
 	unregister_adapter_interrupt(&gib_alert_irq);
 	free_page((unsigned long)gib);
@@ -3315,6 +3321,14 @@ int kvm_s390_gib_init(u8 nisc)
 		goto out_unreg_gal;
 	}
 
+	if (IS_ENABLED(CONFIG_PCI) && sclp.has_aeni) {
+		if (kvm_s390_pci_aen_init(nisc)) {
+			pr_err("Initializing AEN for PCI failed\n");
+			rc = -EIO;
+			goto out_unreg_gal;
+		}
+	}
+
 	KVM_EVENT(3, "gib 0x%pK (nisc=%d) initialized", gib, gib->nisc);
 	goto out;
 
arch/s390/kvm/kvm-s390.c +9 -0
index 14a18ba5ff2c..01dc3f6883d0 100644
@@ -48,6 +48,7 @@
 #include <asm/fpu/api.h>
 #include "kvm-s390.h"
 #include "gaccess.h"
+#include "pci.h"
 
 #define CREATE_TRACE_POINTS
 #include "trace.h"
@@ -503,6 +504,14 @@ int kvm_arch_init(void *opaque)
 		goto out;
 	}
 
+	if (IS_ENABLED(CONFIG_PCI)) {
+		rc = kvm_s390_pci_init();
+		if (rc) {
+			pr_err("Unable to allocate AIFT for PCI\n");
+			goto out;
+		}
+	}
+
 	rc = kvm_s390_gib_init(GAL_ISC);
 	if (rc)
 		goto out;
arch/s390/kvm/pci.c +144 -0

153 lines not on this page · raw message

arch/s390/kvm/pci.h new file +42 -0

42 lines not on this page · raw message

arch/s390/pci/pci.c +6 -0
index 1e939b4cf25e..2a19becbc14c 100644
@@ -61,6 +61,12 @@ DEFINE_STATIC_KEY_FALSE(have_mio);
 
 static struct kmem_cache *zdev_fmb_cache;
 
+/* AEN structures that must be preserved over KVM module re-insertion */
+union zpci_sic_iib *zpci_aipb;
+EXPORT_SYMBOL_GPL(zpci_aipb);
+struct airq_iv *zpci_aif_sbv;
+EXPORT_SYMBOL_GPL(zpci_aif_sbv);
+
 struct zpci_dev *get_zdev_by_fid(u32 fid)
 {
 	struct zpci_dev *tmp, *zdev = NULL;
Signature · 1 line
2.27.0
#39 lore raw

Re: [PATCH v2 15/30] KVM: s390: pci: do initial setup for AEN interpretation

Quoted 190 lines · Matthew Rosato

Initial setup for Adapter Event Notification Interpretation for zPCI passthrough devices. Specifically, allocate a structure for forwarding of adapter events and pass the address of this structure to firmware.

---
  arch/s390/include/asm/pci.h      |   4 +
  arch/s390/include/asm/pci_insn.h |  12 +++
  arch/s390/kvm/interrupt.c        |  14 +++
  arch/s390/kvm/kvm-s390.c         |   9 ++
  arch/s390/kvm/pci.c              | 144 +++++++++++++++++++++++++++++++
  arch/s390/kvm/pci.h              |  42 +++++++++
  arch/s390/pci/pci.c              |   6 ++
  7 files changed, 231 insertions(+)

create mode 100644 arch/s390/kvm/pci.h

arch/s390/include/asm/pci.h +4 -0
index 9b6c657d8d31..9ff8dc19975e 100644
@@ -9,6 +9,7 @@
  #include <asm-generic/pci.h>
  #include <asm/pci_clp.h>
  #include <asm/pci_debug.h>
+#include <asm/pci_insn.h>
  #include <asm/sclp.h>
 
  #define PCIBIOS_MIN_IO		0x1000
@@ -204,6 +205,9 @@ extern const struct attribute_group *zpci_attr_groups[];
  extern unsigned int s390_pci_force_floating __initdata;
  extern unsigned int s390_pci_no_rid;
 
+extern union zpci_sic_iib *zpci_aipb;
+extern struct airq_iv *zpci_aif_sbv;
+
  /* -----------------------------------------------------------------------------
    Prototypes
  ----------------------------------------------------------------------------- */
arch/s390/include/asm/pci_insn.h +12 -0
index 32759c407b8f..ad9000295c82 100644
@@ -101,6 +101,7 @@ struct zpci_fib {
  /* Set Interruption Controls Operation Controls  */
  #define	SIC_IRQ_MODE_ALL		0
  #define	SIC_IRQ_MODE_SINGLE		1
+#define	SIC_SET_AENI_CONTROLS		2
  #define	SIC_IRQ_MODE_DIRECT		4
  #define	SIC_IRQ_MODE_D_ALL		16
  #define	SIC_IRQ_MODE_D_SINGLE		17
@@ -127,9 +128,20 @@ struct zpci_cdiib {
  	u64 : 64;
  } __packed __aligned(8);
 
+/* adapter interruption parameters block */
+struct zpci_aipb {
+	u64 faisb;
+	u64 gait;
+	u16 : 13;
+	u16 afi : 3;
+	u32 : 32;
+	u16 faal;
+} __packed __aligned(8);
+
  union zpci_sic_iib {
  	struct zpci_diib diib;
  	struct zpci_cdiib cdiib;
+	struct zpci_aipb aipb;
  };
 
  DECLARE_STATIC_KEY_FALSE(have_mio);
arch/s390/kvm/interrupt.c +14 -0
index f9b872e358c6..a591b8cd662f 100644
@@ -32,6 +32,7 @@
  #include "kvm-s390.h"
  #include "gaccess.h"
  #include "trace-s390.h"
+#include "pci.h"
 
  #define PFAULT_INIT 0x0600
  #define PFAULT_DONE 0x0680
@@ -3278,6 +3279,11 @@ void kvm_s390_gib_destroy(void)
  {
  	if (!gib)
  		return;
+	if (IS_ENABLED(CONFIG_PCI) && sclp.has_aeni && aift) {
+		mutex_lock(&aift->lock);
+		kvm_s390_pci_aen_exit();
+		mutex_unlock(&aift->lock);
+	}
  	chsc_sgib(0);
  	unregister_adapter_interrupt(&gib_alert_irq);
  	free_page((unsigned long)gib);
@@ -3315,6 +3321,14 @@ int kvm_s390_gib_init(u8 nisc)
  		goto out_unreg_gal;
  	}
 
+	if (IS_ENABLED(CONFIG_PCI) && sclp.has_aeni) {
+		if (kvm_s390_pci_aen_init(nisc)) {
+			pr_err("Initializing AEN for PCI failed\n");
+			rc = -EIO;
+			goto out_unreg_gal;
+		}
+	}
+
  	KVM_EVENT(3, "gib 0x%pK (nisc=%d) initialized", gib, gib->nisc);
  	goto out;
 
arch/s390/kvm/kvm-s390.c +9 -0
index 14a18ba5ff2c..01dc3f6883d0 100644
@@ -48,6 +48,7 @@
  #include <asm/fpu/api.h>
  #include "kvm-s390.h"
  #include "gaccess.h"
+#include "pci.h"
 
  #define CREATE_TRACE_POINTS
  #include "trace.h"
@@ -503,6 +504,14 @@ int kvm_arch_init(void *opaque)
  		goto out;
  	}
 
+	if (IS_ENABLED(CONFIG_PCI)) {
+		rc = kvm_s390_pci_init();
+		if (rc) {
+			pr_err("Unable to allocate AIFT for PCI\n");
+			goto out;
+		}
+	}
+
  	rc = kvm_s390_gib_init(GAL_ISC);
  	if (rc)
  		goto out;
arch/s390/kvm/pci.c +45 -0
index 1c33bc7bf2bd..dae853da6df1 100644
@@ -10,6 +10,138 @@
  #include <linux/kvm_host.h>
  #include <linux/pci.h>
  #include <asm/kvm_pci.h>
+#include <asm/pci.h>
+#include <asm/pci_insn.h>
+#include "pci.h"
+
+struct zpci_aift *aift;
+
+static inline int __set_irq_noiib(u16 ctl, u8 isc)
+{
+	union zpci_sic_iib iib = {{0}};
+
+	return zpci_set_irq_ctrl(ctl, isc, &iib);
+}
+
+/* Caller must hold the aift lock before calling this function */
+void kvm_s390_pci_aen_exit(void)
+{
+	unsigned long flags;
+	struct kvm_zdev **gait_kzdev;
+
+	/*
+	 * Contents of the aipb remain registered for the life of the host
+	 * kernel, the information preserved in zpci_aipb and zpci_aif_sbv
+	 * in case we insert the KVM module again later.  Clear the AIFT
+	 * information and free anything not registered with underlying
+	 * firmware.
+	 */
+	spin_lock_irqsave(&aift->gait_lock, flags);
+	gait_kzdev = aift->kzdev;
+	aift->gait = 0;
+	aift->sbv = 0;
+	aift->kzdev = 0;
+	spin_unlock_irqrestore(&aift->gait_lock, flags);
+
+	kfree(gait_kzdev);
+}
+
+int kvm_s390_pci_aen_init(u8 nisc)
+{
+	struct page *page;
+	int rc = 0, size;
+	bool first = false;
+
+	/* If already enabled for AEN, bail out now */
+	if (aift->gait || aift->sbv)
+		return -EPERM;

arch/s390/kvm/pci.c line 67

Quoted 10 lines
arch/s390/kvm/pci.c +10 -0
+
+	mutex_lock(&aift->lock);
+	aift->kzdev = kcalloc(ZPCI_NR_DEVICES, sizeof(struct kvm_zdev),
+			      GFP_KERNEL);
+	if (!aift->kzdev) {
+		rc = -ENOMEM;
+		goto unlock;
+	}
+
+	if (!zpci_aipb) {

I think you should externalize all this allocation and setup of aipb in a dedicated function zpci_setup_aipb()
from here ----->

Quoted 20 lines
arch/s390/kvm/pci.c +20 -0
+		zpci_aipb = kzalloc(sizeof(union zpci_sic_iib), GFP_KERNEL);
+		if (!zpci_aipb) {
+			rc = -ENOMEM;
+			goto free_zdev;
+		}
+		first = true;
+		aift->sbv = airq_iv_create(ZPCI_NR_DEVICES, AIRQ_IV_ALLOC, 0);
+		if (!aift->sbv) {
+			rc = -ENOMEM;
+			goto free_aipb;
+		}
+		zpci_aif_sbv = aift->sbv;
+		size = get_order(PAGE_ALIGN(ZPCI_NR_DEVICES *
+					    sizeof(struct zpci_gaite)));
+		page = alloc_pages(GFP_KERNEL | __GFP_ZERO, size);
+		if (!page) {
+			rc = -ENOMEM;
+			goto free_sbv;
+		}
+		aift->gait = (struct zpci_gaite *)page_to_phys(page);

arch/s390/kvm/pci.c line 97

Quoted 10 lines
arch/s390/kvm/pci.c +10 -0
+
+		zpci_aipb->aipb.faisb = virt_to_phys(aift->sbv->vector);
+		zpci_aipb->aipb.gait = virt_to_phys(aift->gait);
+		zpci_aipb->aipb.afi = nisc;
+		zpci_aipb->aipb.faal = ZPCI_NR_DEVICES;
+
+		/* Setup Adapter Event Notification Interpretation */
+		if (zpci_set_irq_ctrl(SIC_SET_AENI_CONTROLS, 0, zpci_aipb)) {
+			rc = -EIO;
+			goto free_gait;

to here---->

Quoted 14 lines
arch/s390/kvm/pci.c +14 -0
+		}
+	} else {
+		/*
+		 * AEN registration can only happen once per system boot.  If
+		 * an aipb already exists then AEN was already registered and
+		 * we can re-use the aipb contents.  This can only happen if
+		 * the KVM module was removed and re-inserted.
+		 */
+		if (zpci_aipb->aipb.afi != nisc ||
+		    zpci_aipb->aipb.faal != ZPCI_NR_DEVICES) {
+			rc = -EINVAL;
+			goto free_zdev;
+		}
+		aift->sbv = zpci_aif_sbv;

arch/s390/kvm/pci.c line 121

Quoted 10 lines
arch/s390/kvm/pci.c +10 -0
+		aift->gait = (struct zpci_gaite *)zpci_aipb->aipb.gait;
+	}
+
+	/* Enable floating IRQs */
+	if (__set_irq_noiib(SIC_IRQ_MODE_SINGLE, nisc)) {
+		rc = -EIO;
+		kvm_s390_pci_aen_exit();
+	}
+
+	goto unlock;

and the according errors

here ---->

Quoted 7 lines
arch/s390/kvm/pci.c +7 -0
+
+free_gait:
+	size = get_order(PAGE_ALIGN(ZPCI_NR_DEVICES *
+				    sizeof(struct zpci_gaite)));
+	free_pages((unsigned long)aift->gait, size);
+free_sbv:
+	if (first) {

arch/s390/kvm/pci.c line 138

Quoted 10 lines
arch/s390/kvm/pci.c +10 -0
+		/* If AEN setup failed, only free a newly-allocated sbv */
+		airq_iv_release(aift->sbv);
+		zpci_aif_sbv = 0;
+	}
+free_aipb:
+	if (first) {
+		/* If AEN setup failed, only free a newly-allocated aipb */
+		kfree(zpci_aipb);
+		zpci_aipb = 0;
+	}

to here ---->

To simplify the understanding.

arch/s390/kvm/pci.c line 145

Quoted 7 lines
arch/s390/kvm/pci.c +6 -0
+free_zdev:
+	kfree(aift->kzdev);
+unlock:
+	mutex_unlock(&aift->lock);
+	return rc;
+}
 

... snip...

The second part of the if(aipb) else
could also be externalise in zpci_reset_aipb()

which leads to

if(!aipb)
ret = zpci_setup_aipb()
else
ret = zpci_reset_aipb()

if (ret)

	goto cleanup;

    enable_irq()
	goto unlock;

I think that if we can do that it would be much clearer. what do you think?

Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#40 lore raw

Re: [PATCH v2 15/30] KVM: s390: pci: do initial setup for AEN interpretation

Quoted 307 lines · Pierre Morel
Quoted 186 lines · Matthew Rosato

Initial setup for Adapter Event Notification Interpretation for zPCI passthrough devices.  Specifically, allocate a structure for forwarding of adapter events and pass the address of this structure to firmware.

---
  arch/s390/include/asm/pci.h      |   4 +
  arch/s390/include/asm/pci_insn.h |  12 +++
  arch/s390/kvm/interrupt.c        |  14 +++
  arch/s390/kvm/kvm-s390.c         |   9 ++

  arch/s390/kvm/pci.c              | 144 +++++++++++++++++++++++++++++++
  arch/s390/kvm/pci.h              |  42 +++++++++
  arch/s390/pci/pci.c              |   6 ++
  7 files changed, 231 insertions(+)

  create mode 100644 arch/s390/kvm/pci.h

arch/s390/include/asm/pci.h +4 -1
index 9b6c657d8d31..9ff8dc19975e 100644
@@ -9,6 +9,7 @@
   #include <asm-generic/pci.h>
   #include <asm/pci_clp.h>
   #include <asm/pci_debug.h>
+#include <asm/pci_insn.h>
   #include <asm/sclp.h>
   #define PCIBIOS_MIN_IO        0x1000
@@ -204,6 +205,9 @@ extern const struct attribute_group 
 *zpci_attr_groups[];
   extern unsigned int s390_pci_force_floating __initdata;
   extern unsigned int s390_pci_no_rid;
+extern union zpci_sic_iib *zpci_aipb;
+extern struct airq_iv *zpci_aif_sbv;
+
   /* 
----------------------------------------------------------------------------- 
 

    Prototypes

-----------------------------------------------------------------------------
*/

arch/s390/include/asm/pci_insn.h +12 -0
index 32759c407b8f..ad9000295c82 100644
@@ -101,6 +101,7 @@ struct zpci_fib {
   /* Set Interruption Controls Operation Controls  */
   #define    SIC_IRQ_MODE_ALL        0
   #define    SIC_IRQ_MODE_SINGLE        1
+#define    SIC_SET_AENI_CONTROLS        2
   #define    SIC_IRQ_MODE_DIRECT        4
   #define    SIC_IRQ_MODE_D_ALL        16
   #define    SIC_IRQ_MODE_D_SINGLE        17
@@ -127,9 +128,20 @@ struct zpci_cdiib {
       u64 : 64;
   } __packed __aligned(8);
+/* adapter interruption parameters block */
+struct zpci_aipb {
+    u64 faisb;
+    u64 gait;
+    u16 : 13;
+    u16 afi : 3;
+    u32 : 32;
+    u16 faal;
+} __packed __aligned(8);
+
   union zpci_sic_iib {
       struct zpci_diib diib;
       struct zpci_cdiib cdiib;
+    struct zpci_aipb aipb;
   };
   DECLARE_STATIC_KEY_FALSE(have_mio);
arch/s390/kvm/interrupt.c +14 -0
index f9b872e358c6..a591b8cd662f 100644
@@ -32,6 +32,7 @@
   #include "kvm-s390.h"
   #include "gaccess.h"
   #include "trace-s390.h"
+#include "pci.h"
   #define PFAULT_INIT 0x0600
   #define PFAULT_DONE 0x0680
@@ -3278,6 +3279,11 @@ void kvm_s390_gib_destroy(void)
   {
       if (!gib)
           return;
+    if (IS_ENABLED(CONFIG_PCI) && sclp.has_aeni && aift) {
+        mutex_lock(&aift->lock);
+        kvm_s390_pci_aen_exit();
+        mutex_unlock(&aift->lock);
+    }
       chsc_sgib(0);
       unregister_adapter_interrupt(&gib_alert_irq);
       free_page((unsigned long)gib);
@@ -3315,6 +3321,14 @@ int kvm_s390_gib_init(u8 nisc)
           goto out_unreg_gal;
       }
+    if (IS_ENABLED(CONFIG_PCI) && sclp.has_aeni) {
+        if (kvm_s390_pci_aen_init(nisc)) {
+            pr_err("Initializing AEN for PCI failed\n");
+            rc = -EIO;
+            goto out_unreg_gal;
+        }
+    }
+
       KVM_EVENT(3, "gib 0x%pK (nisc=%d) initialized", gib, gib->nisc);
       goto out;
arch/s390/kvm/kvm-s390.c +9 -0
index 14a18ba5ff2c..01dc3f6883d0 100644
@@ -48,6 +48,7 @@
   #include <asm/fpu/api.h>
   #include "kvm-s390.h"
   #include "gaccess.h"
+#include "pci.h"
   #define CREATE_TRACE_POINTS
   #include "trace.h"
@@ -503,6 +504,14 @@ int kvm_arch_init(void *opaque)
           goto out;
       }
+    if (IS_ENABLED(CONFIG_PCI)) {
+        rc = kvm_s390_pci_init();
+        if (rc) {
+            pr_err("Unable to allocate AIFT for PCI\n");
+            goto out;
+        }
+    }
+
       rc = kvm_s390_gib_init(GAL_ISC);
       if (rc)
           goto out;
arch/s390/kvm/pci.c +45 -0
index 1c33bc7bf2bd..dae853da6df1 100644
@@ -10,6 +10,138 @@
   #include <linux/kvm_host.h>
   #include <linux/pci.h>
   #include <asm/kvm_pci.h>
+#include <asm/pci.h>
+#include <asm/pci_insn.h>
+#include "pci.h"
+
+struct zpci_aift *aift;
+
+static inline int __set_irq_noiib(u16 ctl, u8 isc)
+{
+    union zpci_sic_iib iib = {{0}};
+
+    return zpci_set_irq_ctrl(ctl, isc, &iib);
+}
+
+/* Caller must hold the aift lock before calling this function */
+void kvm_s390_pci_aen_exit(void)
+{
+    unsigned long flags;
+    struct kvm_zdev **gait_kzdev;
+
+    /*
+     * Contents of the aipb remain registered for the life of the host
+     * kernel, the information preserved in zpci_aipb and zpci_aif_sbv
+     * in case we insert the KVM module again later.  Clear the AIFT
+     * information and free anything not registered with underlying
+     * firmware.
+     */
+    spin_lock_irqsave(&aift->gait_lock, flags);
+    gait_kzdev = aift->kzdev;
+    aift->gait = 0;
+    aift->sbv = 0;
+    aift->kzdev = 0;
+    spin_unlock_irqrestore(&aift->gait_lock, flags);
+
+    kfree(gait_kzdev);
+}
+
+int kvm_s390_pci_aen_init(u8 nisc)
+{
+    struct page *page;
+    int rc = 0, size;
+    bool first = false;
+
+    /* If already enabled for AEN, bail out now */
+    if (aift->gait || aift->sbv)
+        return -EPERM;

arch/s390/kvm/pci.c line 67

Quoted 10 lines
arch/s390/kvm/pci.c +10 -0
+
+    mutex_lock(&aift->lock);
+    aift->kzdev = kcalloc(ZPCI_NR_DEVICES, sizeof(struct kvm_zdev),
+                  GFP_KERNEL);
+    if (!aift->kzdev) {
+        rc = -ENOMEM;
+        goto unlock;
+    }
+
+    if (!zpci_aipb) {

I think you should externalize all this allocation and setup of aipb in a dedicated function zpci_setup_aipb()
from here ----->

Quoted 20 lines
arch/s390/kvm/pci.c +20 -0
+        zpci_aipb = kzalloc(sizeof(union zpci_sic_iib), GFP_KERNEL);
+        if (!zpci_aipb) {
+            rc = -ENOMEM;
+            goto free_zdev;
+        }
+        first = true;
+        aift->sbv = airq_iv_create(ZPCI_NR_DEVICES, AIRQ_IV_ALLOC, 0);
+        if (!aift->sbv) {
+            rc = -ENOMEM;
+            goto free_aipb;
+        }
+        zpci_aif_sbv = aift->sbv;
+        size = get_order(PAGE_ALIGN(ZPCI_NR_DEVICES *
+                        sizeof(struct zpci_gaite)));
+        page = alloc_pages(GFP_KERNEL | __GFP_ZERO, size);
+        if (!page) {
+            rc = -ENOMEM;
+            goto free_sbv;
+        }
+        aift->gait = (struct zpci_gaite *)page_to_phys(page);

arch/s390/kvm/pci.c line 97

Quoted 10 lines
arch/s390/kvm/pci.c +10 -0
+
+        zpci_aipb->aipb.faisb = virt_to_phys(aift->sbv->vector);
+        zpci_aipb->aipb.gait = virt_to_phys(aift->gait);
+        zpci_aipb->aipb.afi = nisc;
+        zpci_aipb->aipb.faal = ZPCI_NR_DEVICES;
+
+        /* Setup Adapter Event Notification Interpretation */
+        if (zpci_set_irq_ctrl(SIC_SET_AENI_CONTROLS, 0, zpci_aipb)) {
+            rc = -EIO;
+            goto free_gait;

to here---->

Quoted 14 lines
arch/s390/kvm/pci.c +14 -0
+        }
+    } else {
+        /*
+         * AEN registration can only happen once per system boot.  If
+         * an aipb already exists then AEN was already registered and
+         * we can re-use the aipb contents.  This can only happen if
+         * the KVM module was removed and re-inserted.
+         */
+        if (zpci_aipb->aipb.afi != nisc ||
+            zpci_aipb->aipb.faal != ZPCI_NR_DEVICES) {
+            rc = -EINVAL;
+            goto free_zdev;
+        }
+        aift->sbv = zpci_aif_sbv;

arch/s390/kvm/pci.c line 121

Quoted 10 lines
arch/s390/kvm/pci.c +10 -0
+        aift->gait = (struct zpci_gaite *)zpci_aipb->aipb.gait;
+    }
+
+    /* Enable floating IRQs */
+    if (__set_irq_noiib(SIC_IRQ_MODE_SINGLE, nisc)) {
+        rc = -EIO;
+        kvm_s390_pci_aen_exit();
+    }
+
+    goto unlock;

and the according errors

here ---->

Quoted 7 lines
arch/s390/kvm/pci.c +7 -0
+
+free_gait:
+    size = get_order(PAGE_ALIGN(ZPCI_NR_DEVICES *
+                    sizeof(struct zpci_gaite)));
+    free_pages((unsigned long)aift->gait, size);
+free_sbv:
+    if (first) {

arch/s390/kvm/pci.c line 138

Quoted 10 lines
arch/s390/kvm/pci.c +10 -0
+        /* If AEN setup failed, only free a newly-allocated sbv */
+        airq_iv_release(aift->sbv);
+        zpci_aif_sbv = 0;
+    }
+free_aipb:
+    if (first) {
+        /* If AEN setup failed, only free a newly-allocated aipb */
+        kfree(zpci_aipb);
+        zpci_aipb = 0;
+    }

to here ---->

To simplify the understanding.

arch/s390/kvm/pci.c line 145

Quoted 7 lines
arch/s390/kvm/pci.c +6 -0
+free_zdev:
+    kfree(aift->kzdev);
+unlock:
+    mutex_unlock(&aift->lock);
+    return rc;
+}
 

... snip...

The second part of the if(aipb) else
could also be externalise in zpci_reset_aipb()

which leads to

   if(!aipb)
    ret = zpci_setup_aipb()
   else
    ret = zpci_reset_aipb()

Quoted 9 lines

   if (ret)
    goto cleanup;

    enable_irq()
    goto unlock;

I think that if we can do that it would be much clearer. what do you think?

Yup, that sounds good, I will re-organize with 2 new static functions zpci_setup_aipb() and zpci_reset_aipb()

#41 lore raw

Re: [PATCH v2 15/30] KVM: s390: pci: do initial setup for AEN interpretation

In reply to Matthew Rosato
Quoted 16 lines · Matthew Rosato

Initial setup for Adapter Event Notification Interpretation for zPCI passthrough devices. Specifically, allocate a structure for forwarding of adapter events and pass the address of this structure to firmware.

---
  arch/s390/include/asm/pci.h      |   4 +
  arch/s390/include/asm/pci_insn.h |  12 +++
  arch/s390/kvm/interrupt.c        |  14 +++
  arch/s390/kvm/kvm-s390.c         |   9 ++
  arch/s390/kvm/pci.c              | 144 +++++++++++++++++++++++++++++++
  arch/s390/kvm/pci.h              |  42 +++++++++
  arch/s390/pci/pci.c              |   6 ++
  7 files changed, 231 insertions(+)

create mode 100644 arch/s390/kvm/pci.h

...snip...

Quoted 27 lines

new file mode 100644
index 000000000000..b2000ed7b8c3

arch/s390/kvm/pci.h new file +22 -0
@@ -0,0 +1,42 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * s390 kvm PCI passthrough support
+ *
+ * Copyright IBM Corp. 2021
+ *
+ *    Author(s): Matthew Rosato <[email protected]>
+ */
+
+#ifndef __KVM_S390_PCI_H
+#define __KVM_S390_PCI_H
+
+#include <linux/pci.h>
+#include <linux/mutex.h>
+#include <asm/airq.h>
+#include <asm/kvm_pci.h>
+
+struct zpci_gaite {
+	u32 gisa;
+	u8 gisc;
+	u8 count;
+	u8 reserved;

arch/s390/kvm/pci.h line 32

Quoted 10 lines
arch/s390/kvm/pci.h new file +10 -0
+	u8 aisbo;
+	u64 aisb;
+};
+
+struct zpci_aift {
+	struct zpci_gaite *gait;
+	struct airq_iv *sbv;
+	struct kvm_zdev **kzdev;
+	spinlock_t gait_lock; /* Protects the gait, used during AEN forward */
+	struct mutex lock; /* Protects the other structures in aift */

To facilitate review and debug, can we please rename the lock aift_lock?

Quoted 4 lines
arch/s390/kvm/pci.h new file +4 -0
+};
+
+extern struct zpci_aift *aift;
+

...snip...

Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#42 lore raw

Re: [PATCH v2 15/30] KVM: s390: pci: do initial setup for AEN interpretation

Quoted 53 lines · Pierre Morel
Quoted 16 lines · Matthew Rosato

Initial setup for Adapter Event Notification Interpretation for zPCI passthrough devices.  Specifically, allocate a structure for forwarding of adapter events and pass the address of this structure to firmware.

---
  arch/s390/include/asm/pci.h      |   4 +
  arch/s390/include/asm/pci_insn.h |  12 +++
  arch/s390/kvm/interrupt.c        |  14 +++
  arch/s390/kvm/kvm-s390.c         |   9 ++

  arch/s390/kvm/pci.c              | 144 +++++++++++++++++++++++++++++++
  arch/s390/kvm/pci.h              |  42 +++++++++
  arch/s390/pci/pci.c              |   6 ++
  7 files changed, 231 insertions(+)

  create mode 100644 arch/s390/kvm/pci.h

...snip...

Quoted 32 lines

new file mode 100644
index 000000000000..b2000ed7b8c3

arch/s390/kvm/pci.h new file +27 -0
@@ -0,0 +1,42 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * s390 kvm PCI passthrough support
+ *
+ * Copyright IBM Corp. 2021
+ *
+ *    Author(s): Matthew Rosato <[email protected]>
+ */
+
+#ifndef __KVM_S390_PCI_H
+#define __KVM_S390_PCI_H
+
+#include <linux/pci.h>
+#include <linux/mutex.h>
+#include <asm/airq.h>
+#include <asm/kvm_pci.h>
+
+struct zpci_gaite {
+    u32 gisa;
+    u8 gisc;
+    u8 count;
+    u8 reserved;
+    u8 aisbo;
+    u64 aisb;
+};
+
+struct zpci_aift {
Quoted 10 lines
Quoted 6 lines
arch/s390/kvm/pci.h new file +4 -0
+    struct zpci_gaite *gait;
+    struct airq_iv *sbv;
+    struct kvm_zdev **kzdev;
+    spinlock_t gait_lock; /* Protects the gait, used during AEN 

forward */
+    struct mutex lock; /* Protects the other structures in aift */

To facilitate review and debug, can we please rename the lock aift_lock?

OK, sure

author maintainer
#43 lore raw

PATCH v2 16/30 KVM: s390: pci: enable host forwarding of Adapter Event Notifications

In reply to Matthew Rosato

In cases where interrupts are not forwarded to the guest via firmware, KVM is responsible for ensuring delivery. When an interrupt presents with the forwarding bit, we must process the forwarding tables until all interrupts are delivered.

---
 arch/s390/include/asm/kvm_host.h |  1 +
 arch/s390/include/asm/tpi.h      | 13 ++++++
 arch/s390/kvm/interrupt.c        | 76 +++++++++++++++++++++++++++++++-
 arch/s390/kvm/kvm-s390.c         |  3 +-
 arch/s390/kvm/pci.h              |  9 ++++
 5 files changed, 100 insertions(+), 2 deletions(-)
arch/s390/include/asm/kvm_host.h +1 -0
index a604d51acfc8..3f147b8d050b 100644
@@ -757,6 +757,7 @@ struct kvm_vm_stat {
 	u64 inject_pfault_done;
 	u64 inject_service_signal;
 	u64 inject_virtio;
+	u64 aen_forward;
 };
 
 struct kvm_arch_memory_slot {
arch/s390/include/asm/tpi.h +13 -0
index 1ac538b8cbf5..f76e5fdff23a 100644
@@ -19,6 +19,19 @@ struct tpi_info {
 	u32 :12;
 } __packed __aligned(4);
 
+/* I/O-Interruption Code as stored by TPI for an Adapter I/O */
+struct tpi_adapter_info {
+	u32 aism:8;
+	u32 :22;
+	u32 error:1;
+	u32 forward:1;
+	u32 reserved;
+	u32 adapter_IO:1;
+	u32 directed_irq:1;
+	u32 isc:3;
+	u32 :27;
+} __packed __aligned(4);
+
 #endif /* __ASSEMBLY__ */
 
 #endif /* _ASM_S390_TPI_H */
arch/s390/kvm/interrupt.c +75 -1

86 lines not on this page · raw message

arch/s390/kvm/kvm-s390.c +2 -1
index 01dc3f6883d0..ab8b56deed11 100644
@@ -65,7 +65,8 @@ const struct _kvm_stats_desc kvm_vm_stats_desc[] = {
 	STATS_DESC_COUNTER(VM, inject_float_mchk),
 	STATS_DESC_COUNTER(VM, inject_pfault_done),
 	STATS_DESC_COUNTER(VM, inject_service_signal),
-	STATS_DESC_COUNTER(VM, inject_virtio)
+	STATS_DESC_COUNTER(VM, inject_virtio),
+	STATS_DESC_COUNTER(VM, aen_forward)
 };
 
 const struct kvm_stats_header kvm_vm_stats_header = {
arch/s390/kvm/pci.h +9 -0
index b2000ed7b8c3..387b637863c9 100644
@@ -12,6 +12,7 @@
 
 #include <linux/pci.h>
 #include <linux/mutex.h>
+#include <linux/kvm_host.h>
 #include <asm/airq.h>
 #include <asm/kvm_pci.h>
 
@@ -34,6 +35,14 @@ struct zpci_aift {
 
 extern struct zpci_aift *aift;
 
+static inline struct kvm *kvm_s390_pci_si_to_kvm(struct zpci_aift *aift,
+						 unsigned long si)
+{
+	if (!IS_ENABLED(CONFIG_PCI) || aift->kzdev == 0 || aift->kzdev[si] == 0)
+		return 0;
+	return aift->kzdev[si]->kvm;
+};
+
 int kvm_s390_pci_aen_init(u8 nisc);
 void kvm_s390_pci_aen_exit(void);
 
Signature · 1 line
2.27.0
#44 lore raw

Re: [PATCH v2 16/30] KVM: s390: pci: enable host forwarding of Adapter Event Notifications

Quoted 98 lines · Matthew Rosato

In cases where interrupts are not forwarded to the guest via firmware, KVM is responsible for ensuring delivery. When an interrupt presents with the forwarding bit, we must process the forwarding tables until all interrupts are delivered.

---
  arch/s390/include/asm/kvm_host.h |  1 +
  arch/s390/include/asm/tpi.h      | 13 ++++++
  arch/s390/kvm/interrupt.c        | 76 +++++++++++++++++++++++++++++++-
  arch/s390/kvm/kvm-s390.c         |  3 +-
  arch/s390/kvm/pci.h              |  9 ++++
  5 files changed, 100 insertions(+), 2 deletions(-)
arch/s390/include/asm/kvm_host.h +1 -0
index a604d51acfc8..3f147b8d050b 100644
@@ -757,6 +757,7 @@ struct kvm_vm_stat {
  	u64 inject_pfault_done;
  	u64 inject_service_signal;
  	u64 inject_virtio;
+	u64 aen_forward;
  };
 
  struct kvm_arch_memory_slot {
arch/s390/include/asm/tpi.h +13 -0
index 1ac538b8cbf5..f76e5fdff23a 100644
@@ -19,6 +19,19 @@ struct tpi_info {
  	u32 :12;
  } __packed __aligned(4);
 
+/* I/O-Interruption Code as stored by TPI for an Adapter I/O */
+struct tpi_adapter_info {
+	u32 aism:8;
+	u32 :22;
+	u32 error:1;
+	u32 forward:1;
+	u32 reserved;
+	u32 adapter_IO:1;
+	u32 directed_irq:1;
+	u32 isc:3;
+	u32 :27;
+} __packed __aligned(4);
+
  #endif /* __ASSEMBLY__ */
 
  #endif /* _ASM_S390_TPI_H */
arch/s390/kvm/interrupt.c +40 -0
index a591b8cd662f..07743c6a67c4 100644
@@ -3263,11 +3263,85 @@ int kvm_s390_gisc_unregister(struct kvm *kvm, u32 gisc)
  }
  EXPORT_SYMBOL_GPL(kvm_s390_gisc_unregister);
 
+static void aen_host_forward(unsigned long si)
+{
+	struct kvm_s390_gisa_interrupt *gi;
+	struct zpci_gaite *gaite;
+	struct kvm *kvm;
+
+	gaite = (struct zpci_gaite *)aift->gait +
+		(si * sizeof(struct zpci_gaite));
+	if (gaite->count == 0)
+		return;
+	if (gaite->aisb != 0)
+		set_bit_inv(gaite->aisbo, (unsigned long *)gaite->aisb);
+
+	kvm = kvm_s390_pci_si_to_kvm(aift, si);
+	if (kvm == 0)
+		return;
+	gi = &kvm->arch.gisa_int;
+
+	if (!(gi->origin->g1.simm & AIS_MODE_MASK(gaite->gisc)) ||
+	    !(gi->origin->g1.nimm & AIS_MODE_MASK(gaite->gisc))) {
+		gisa_set_ipm_gisc(gi->origin, gaite->gisc);
+		if (hrtimer_active(&gi->timer))
+			hrtimer_cancel(&gi->timer);
+		hrtimer_start(&gi->timer, 0, HRTIMER_MODE_REL);
+		kvm->stat.aen_forward++;
+	}
+}
+
+static void aen_process_gait(u8 isc)
+{
+	bool found = false, first = true;
+	union zpci_sic_iib iib = {{0}};
+	unsigned long si, flags;
+
+	spin_lock_irqsave(&aift->gait_lock, flags);
+
+	if (!aift->gait) {
+		spin_unlock_irqrestore(&aift->gait_lock, flags);
+		return;
+	}

arch/s390/kvm/interrupt.c int kvm_s390_gisc_unregister(struct kvm *kvm, u32 gisc) · line 3315

Quoted 10 lines
arch/s390/kvm/interrupt.c +10 -0
+
+	for (si = 0;;) {
+		/* Scan adapter summary indicator bit vector */
+		si = airq_iv_scan(aift->sbv, si, airq_iv_end(aift->sbv));
+		if (si == -1UL) {
+			if (first || found) {
+				/* Reenable interrupts. */
+				if (zpci_set_irq_ctrl(SIC_IRQ_MODE_SINGLE, isc,
+						      &iib))
+					break;

AFAIU this code is VFIO interpretation specific code and facility 12 is a precondition for it, so I think this break will never occur. If I am right we should not test the return value which will make the code clearer.

arch/s390/kvm/interrupt.c int kvm_s390_gisc_unregister(struct kvm *kvm, u32 gisc) · line 3320

Quoted 5 lines
arch/s390/kvm/interrupt.c +5 -0
+				first = found = false;
+			} else {
+				/* Interrupts on and all bits processed */
+				break;
+			}

May be add a comment: "rescan after re-enabling interrupts"

Quoted 15 lines
arch/s390/kvm/interrupt.c +12 -0
+			found = false;
+			si = 0;
+			continue;
+		}
+		found = true;
+		aen_host_forward(si);
+	}
+
+	spin_unlock_irqrestore(&aift->gait_lock, flags);
+}
+
  static void gib_alert_irq_handler(struct airq_struct *airq,
  				  struct tpi_info *tpi_info)
  {
+	struct tpi_adapter_info *info = (struct tpi_adapter_info *)tpi_info;

arch/s390/kvm/interrupt.c int kvm_s390_gisc_unregister(struct kvm *kvm, u32 gisc) · line 3344

Quoted 10 lines
arch/s390/kvm/interrupt.c +8 -1
+
  	inc_irq_stat(IRQIO_GAL);
-	process_gib_alert_list();
+
+	if (IS_ENABLED(CONFIG_PCI) && (info->forward || info->error)) {
+		aen_process_gait(info->isc);
+		if (info->aism != 0)
+			process_gib_alert_list();
+	} else
+		process_gib_alert_list();

NIT: I think we need braces around this statement

Quoted 30 lines
arch/s390/kvm/interrupt.c +0 -0
  }
 
  static struct airq_struct gib_alert_irq = {
arch/s390/kvm/kvm-s390.c +2 -1
index 01dc3f6883d0..ab8b56deed11 100644
@@ -65,7 +65,8 @@ const struct _kvm_stats_desc kvm_vm_stats_desc[] = {
  	STATS_DESC_COUNTER(VM, inject_float_mchk),
  	STATS_DESC_COUNTER(VM, inject_pfault_done),
  	STATS_DESC_COUNTER(VM, inject_service_signal),
-	STATS_DESC_COUNTER(VM, inject_virtio)
+	STATS_DESC_COUNTER(VM, inject_virtio),
+	STATS_DESC_COUNTER(VM, aen_forward)
  };
 
  const struct kvm_stats_header kvm_vm_stats_header = {
arch/s390/kvm/pci.h +1 -0
index b2000ed7b8c3..387b637863c9 100644
@@ -12,6 +12,7 @@
 
  #include <linux/pci.h>
  #include <linux/mutex.h>
+#include <linux/kvm_host.h>
  #include <asm/airq.h>
  #include <asm/kvm_pci.h>
 

arch/s390/kvm/pci.h struct zpci_aift { · line 41

Quoted 7 lines
arch/s390/kvm/pci.h +4 -0
@@ -34,6 +35,14 @@ struct zpci_aift {
 
  extern struct zpci_aift *aift;
 
+static inline struct kvm *kvm_s390_pci_si_to_kvm(struct zpci_aift *aift,
+						 unsigned long si)
+{
+	if (!IS_ENABLED(CONFIG_PCI) || aift->kzdev == 0 || aift->kzdev[si] == 0)

Shouldn't it be better CONFIG_VFIO_PCI ?

Quoted 8 lines
arch/s390/kvm/pci.h +4 -0
+		return 0;
+	return aift->kzdev[si]->kvm;
+};
+
  int kvm_s390_pci_aen_init(u8 nisc);
  void kvm_s390_pci_aen_exit(void);
 
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#45 lore raw

Re: [PATCH v2 16/30] KVM: s390: pci: enable host forwarding of Adapter Event Notifications

Quoted 1 line · Pierre Morel

...

Quoted 17 lines
Quoted 17 lines
arch/s390/kvm/interrupt.c +17 -0
+static void aen_process_gait(u8 isc)
+{
+    bool found = false, first = true;
+    union zpci_sic_iib iib = {{0}};
+    unsigned long si, flags;
+
+    spin_lock_irqsave(&aift->gait_lock, flags);
+
+    if (!aift->gait) {
+        spin_unlock_irqrestore(&aift->gait_lock, flags);
+        return;
+    }
+
+    for (si = 0;;) {
+        /* Scan adapter summary indicator bit vector */
+        si = airq_iv_scan(aift->sbv, si, airq_iv_end(aift->sbv));
+        if (si == -1UL) {
Quoted 8 lines
Quoted 5 lines

+            if (first || found) {
+                /* Reenable interrupts. */
+                if (zpci_set_irq_ctrl(SIC_IRQ_MODE_SINGLE, isc, +                              &iib))
+                    break;

AFAIU this code is VFIO interpretation specific code and facility 12 is a precondition for it, so I think this break will never occur. If I am right we should not test the return value which will make the code clearer.

Yep, you are correct; we can just ignore the return value here.

Quoted 8 lines
Quoted 6 lines

+                first = found = false;
+            } else {
+                /* Interrupts on and all bits processed */ +                break;
+            }

May be add a comment: "rescan after re-enabling interrupts"

OK

Quoted 20 lines
Quoted 20 lines

+            found = false;
+            si = 0;
+            continue;
+        }
+        found = true;
+        aen_host_forward(si);
+    }

arch/s390/kvm/interrupt.c +4 -0
+
+    spin_unlock_irqrestore(&aift->gait_lock, flags);
+}
+

  static void gib_alert_irq_handler(struct airq_struct *airq, struct tpi_info *tpi_info)
  {
+    struct tpi_adapter_info *info = (struct tpi_adapter_info *)tpi_info; +
      inc_irq_stat(IRQIO_GAL);
-    process_gib_alert_list();
+

Quoted 8 lines
Quoted 6 lines

+    if (IS_ENABLED(CONFIG_PCI) && (info->forward || info->error)) { +        aen_process_gait(info->isc);
+        if (info->aism != 0)
+            process_gib_alert_list();
+    } else
+        process_gib_alert_list();

NIT: I think we need braces around this statement

OK

Quoted 27 lines
Quoted 27 lines

  }
  static struct airq_struct gib_alert_irq = {

arch/s390/kvm/kvm-s390.c +2 -1
index 01dc3f6883d0..ab8b56deed11 100644
@@ -65,7 +65,8 @@ const struct _kvm_stats_desc kvm_vm_stats_desc[] = {
       STATS_DESC_COUNTER(VM, inject_float_mchk),
       STATS_DESC_COUNTER(VM, inject_pfault_done),
       STATS_DESC_COUNTER(VM, inject_service_signal),
-    STATS_DESC_COUNTER(VM, inject_virtio)
+    STATS_DESC_COUNTER(VM, inject_virtio),
+    STATS_DESC_COUNTER(VM, aen_forward)
   };
   const struct kvm_stats_header kvm_vm_stats_header = {
arch/s390/kvm/pci.h +1 -0
index b2000ed7b8c3..387b637863c9 100644
@@ -12,6 +12,7 @@
   #include <linux/pci.h>
   #include <linux/mutex.h>
+#include <linux/kvm_host.h>
   #include <asm/airq.h>
   #include <asm/kvm_pci.h>
@@ -34,6 +35,14 @@ struct zpci_aift {
Quoted 8 lines

arch/s390/kvm/pci.h struct zpci_aift {

Quoted 6 lines
arch/s390/kvm/pci.h +4 -0
   extern struct zpci_aift *aift;
+static inline struct kvm *kvm_s390_pci_si_to_kvm(struct zpci_aift *aift,
+                         unsigned long si)
+{
+    if (!IS_ENABLED(CONFIG_PCI) || aift->kzdev == 0 || 
 aift->kzdev[si] == 0)

Shouldn't it be better CONFIG_VFIO_PCI ?

While it's true that we can't be doing interpretation without CONFIG_VFIO_PCI=y|m, the reason I'm using CONFIG_PCI here and elsewhere in the code is because CONFIG_PCI is what is being used to determine whether or not we build arch/s390/kvm/pci.o in patch 14 (and thus whether or not the aift exists) -- And the reason we use this is because this is where the code dependencies exist (examples include ZPCI_NR_DEVICES, the AEN pieces that must be preserved over KVM module remove/insert in patch 15)

If we for some reason have a case where CONFIG_KVM=y|m && CONFIG_PCI=y|m && CONFIG_VFIO_PCI=n, this will still work: aift and aift->kzdev will exist (kvm/pci.o is linked) but we will never actually drive this routine anyway because we'll never register a device for AEN forwarding without CONFIG_VFIO_PCI.

author maintainer
#46 lore raw

PATCH v2 17/30 KVM: s390: mechanism to enable guest zPCI Interpretation

In reply to Matthew Rosato

The guest must have access to certain facilities in order to allow interpretive execution of zPCI instructions and adapter event notifications. However, there are some cases where a guest might disable interpretation -- provide a mechanism via which we can defer enabling the associated zPCI interpretation facilities until the guest indicates it wishes to use them.

---
 arch/s390/include/asm/kvm_host.h |  4 ++++
 arch/s390/kvm/kvm-s390.c         | 40 ++++++++++++++++++++++++++++++++
 arch/s390/kvm/kvm-s390.h         | 10 ++++++++
 3 files changed, 54 insertions(+)
arch/s390/include/asm/kvm_host.h +4 -0
index 3f147b8d050b..38982c1de413 100644
@@ -252,7 +252,10 @@ struct kvm_s390_sie_block {
 #define ECB2_IEP	0x20
 #define ECB2_PFMFI	0x08
 #define ECB2_ESCA	0x04
+#define ECB2_ZPCI_LSI	0x02
 	__u8    ecb2;                   /* 0x0062 */
+#define ECB3_AISI	0x20
+#define ECB3_AISII	0x10
 #define ECB3_DEA 0x08
 #define ECB3_AES 0x04
 #define ECB3_RI  0x01
@@ -938,6 +941,7 @@ struct kvm_arch{
 	int use_cmma;
 	int use_pfmfi;
 	int use_skf;
+	int use_zpci_interp;
 	int user_cpu_state_ctrl;
 	int user_sigp;
 	int user_stsi;
arch/s390/kvm/kvm-s390.c +40 -0
index ab8b56deed11..b6c32fc3b272 100644
@@ -1029,6 +1029,44 @@ static int kvm_s390_vm_set_crypto(struct kvm *kvm, struct kvm_device_attr *attr)
 	return 0;
 }
 
+static void kvm_s390_vcpu_pci_setup(struct kvm_vcpu *vcpu)
+{
+	/* Only set the ECB bits after guest requests zPCI interpretation */
+	if (!vcpu->kvm->arch.use_zpci_interp)
+		return;
+
+	vcpu->arch.sie_block->ecb2 |= ECB2_ZPCI_LSI;
+	vcpu->arch.sie_block->ecb3 |= ECB3_AISII + ECB3_AISI;
+}
+
+void kvm_s390_vcpu_pci_enable_interp(struct kvm *kvm)
+{
+	struct kvm_vcpu *vcpu;
+	int i;
+
+	/*
+	 * If host is configured for PCI and the necessary facilities are
+	 * available, turn on interpretation for the life of this guest
+	 */
+	if (!IS_ENABLED(CONFIG_PCI) || !sclp.has_zpci_lsi || !sclp.has_aisii ||
+	    !sclp.has_aeni || !sclp.has_aisi)
+		return;
+
+	mutex_lock(&kvm->lock);
+
+	kvm->arch.use_zpci_interp = 1;
+
+	kvm_s390_vcpu_block_all(kvm);
+
+	kvm_for_each_vcpu(i, vcpu, kvm) {
+		kvm_s390_vcpu_pci_setup(vcpu);
+		kvm_s390_sync_request(KVM_REQ_VSIE_RESTART, vcpu);
+	}
+
+	kvm_s390_vcpu_unblock_all(kvm);
+	mutex_unlock(&kvm->lock);
+}
+
 static void kvm_s390_sync_request_broadcast(struct kvm *kvm, int req)
 {
 	int cx;
@@ -3282,6 +3320,8 @@ static int kvm_s390_vcpu_setup(struct kvm_vcpu *vcpu)
 
 	kvm_s390_vcpu_crypto_setup(vcpu);
 
+	kvm_s390_vcpu_pci_setup(vcpu);
+
 	mutex_lock(&vcpu->kvm->lock);
 	if (kvm_s390_pv_is_protected(vcpu->kvm)) {
 		rc = kvm_s390_pv_create_cpu(vcpu, &uvrc, &uvrrc);
arch/s390/kvm/kvm-s390.h +10 -0
index c07a050d757d..a2eccb8b977e 100644
@@ -481,6 +481,16 @@ void kvm_s390_reinject_machine_check(struct kvm_vcpu *vcpu,
  */
 void kvm_s390_vcpu_crypto_reset_all(struct kvm *kvm);
 
+/**
+ * kvm_s390_vcpu_pci_enable_interp
+ *
+ * Set the associated PCI attributes for each vcpu to allow for zPCI Load/Store
+ * interpretation as well as adapter interruption forwarding.
+ *
+ * @kvm: the KVM guest
+ */
+void kvm_s390_vcpu_pci_enable_interp(struct kvm *kvm);
+
 /**
  * diag9c_forwarding_hz
  *
Signature · 1 line
2.27.0
#47 lore raw

Re: [PATCH v2 17/30] KVM: s390: mechanism to enable guest zPCI Interpretation

Quoted 43 lines · Matthew Rosato

The guest must have access to certain facilities in order to allow interpretive execution of zPCI instructions and adapter event notifications. However, there are some cases where a guest might disable interpretation -- provide a mechanism via which we can defer enabling the associated zPCI interpretation facilities until the guest indicates it wishes to use them.

---
  arch/s390/include/asm/kvm_host.h |  4 ++++
  arch/s390/kvm/kvm-s390.c         | 40 ++++++++++++++++++++++++++++++++
  arch/s390/kvm/kvm-s390.h         | 10 ++++++++
  3 files changed, 54 insertions(+)
arch/s390/include/asm/kvm_host.h +4 -0
index 3f147b8d050b..38982c1de413 100644
@@ -252,7 +252,10 @@ struct kvm_s390_sie_block {
  #define ECB2_IEP	0x20
  #define ECB2_PFMFI	0x08
  #define ECB2_ESCA	0x04
+#define ECB2_ZPCI_LSI	0x02
  	__u8    ecb2;                   /* 0x0062 */
+#define ECB3_AISI	0x20
+#define ECB3_AISII	0x10
  #define ECB3_DEA 0x08
  #define ECB3_AES 0x04
  #define ECB3_RI  0x01
@@ -938,6 +941,7 @@ struct kvm_arch{
  	int use_cmma;
  	int use_pfmfi;
  	int use_skf;
+	int use_zpci_interp;
  	int user_cpu_state_ctrl;
  	int user_sigp;
  	int user_stsi;
arch/s390/kvm/kvm-s390.c +0 -0
index ab8b56deed11..b6c32fc3b272 100644
@@ -1029,6 +1029,44 @@ static int kvm_s390_vm_set_crypto(struct kvm *kvm, struct kvm_device_attr *attr)
  	return 0;

arch/s390/kvm/kvm-s390.c static int kvm_s390_vm_set_crypto(struct kvm *kvm, struct kvm_device_attr *attr) · line 1039

Quoted 10 lines
arch/s390/kvm/kvm-s390.c +8 -0
  }
 
+static void kvm_s390_vcpu_pci_setup(struct kvm_vcpu *vcpu)
+{
+	/* Only set the ECB bits after guest requests zPCI interpretation */
+	if (!vcpu->kvm->arch.use_zpci_interp)
+		return;
+
+	vcpu->arch.sie_block->ecb2 |= ECB2_ZPCI_LSI;
+	vcpu->arch.sie_block->ecb3 |= ECB3_AISII + ECB3_AISI;

As far as I understood, the interpretation is only possible if a gisa designation is associated with the PCI function via CLP enable.

Why do we setup the SIE ECB only when the guest requests for interpretation and not systematically in vcpu_setup?

If ECB2_ZPCI_LSI, ECB3_AISII or ECB3_AISI have an effect when the gisa designation is not specified shouldn't we have a way to clear these bits?

Quoted 64 lines
arch/s390/kvm/kvm-s390.c +32 -0
+}
+
+void kvm_s390_vcpu_pci_enable_interp(struct kvm *kvm)
+{
+	struct kvm_vcpu *vcpu;
+	int i;
+
+	/*
+	 * If host is configured for PCI and the necessary facilities are
+	 * available, turn on interpretation for the life of this guest
+	 */
+	if (!IS_ENABLED(CONFIG_PCI) || !sclp.has_zpci_lsi || !sclp.has_aisii ||
+	    !sclp.has_aeni || !sclp.has_aisi)
+		return;
+
+	mutex_lock(&kvm->lock);
+
+	kvm->arch.use_zpci_interp = 1;
+
+	kvm_s390_vcpu_block_all(kvm);
+
+	kvm_for_each_vcpu(i, vcpu, kvm) {
+		kvm_s390_vcpu_pci_setup(vcpu);
+		kvm_s390_sync_request(KVM_REQ_VSIE_RESTART, vcpu);
+	}
+
+	kvm_s390_vcpu_unblock_all(kvm);
+	mutex_unlock(&kvm->lock);
+}
+
  static void kvm_s390_sync_request_broadcast(struct kvm *kvm, int req)
  {
  	int cx;
@@ -3282,6 +3320,8 @@ static int kvm_s390_vcpu_setup(struct kvm_vcpu *vcpu)
 
  	kvm_s390_vcpu_crypto_setup(vcpu);
 
+	kvm_s390_vcpu_pci_setup(vcpu);
+
  	mutex_lock(&vcpu->kvm->lock);
  	if (kvm_s390_pv_is_protected(vcpu->kvm)) {
  		rc = kvm_s390_pv_create_cpu(vcpu, &uvrc, &uvrrc);
arch/s390/kvm/kvm-s390.h +10 -0
index c07a050d757d..a2eccb8b977e 100644
@@ -481,6 +481,16 @@ void kvm_s390_reinject_machine_check(struct kvm_vcpu *vcpu,
   */
  void kvm_s390_vcpu_crypto_reset_all(struct kvm *kvm);
 
+/**
+ * kvm_s390_vcpu_pci_enable_interp
+ *
+ * Set the associated PCI attributes for each vcpu to allow for zPCI Load/Store
+ * interpretation as well as adapter interruption forwarding.
+ *
+ * @kvm: the KVM guest
+ */
+void kvm_s390_vcpu_pci_enable_interp(struct kvm *kvm);
+
  /**
   * diag9c_forwarding_hz
   *
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#48 lore raw

Re: [PATCH v2 17/30] KVM: s390: mechanism to enable guest zPCI Interpretation

Quoted 50 lines · Pierre Morel
Quoted 47 lines · Matthew Rosato

The guest must have access to certain facilities in order to allow interpretive execution of zPCI instructions and adapter event notifications.  However, there are some cases where a guest might disable interpretation -- provide a mechanism via which we can defer enabling the associated zPCI interpretation facilities until the guest indicates it wishes to use them.

---
  arch/s390/include/asm/kvm_host.h |  4 ++++ arch/s390/kvm/kvm-s390.c         | 40 ++++++++++++++++++++++++++++++++ arch/s390/kvm/kvm-s390.h         | 10 ++++++++

  3 files changed, 54 insertions(+)
arch/s390/include/asm/kvm_host.h +1 -0
index 3f147b8d050b..38982c1de413 100644
@@ -252,7 +252,10 @@ struct kvm_s390_sie_block {
   #define ECB2_IEP    0x20
   #define ECB2_PFMFI    0x08
   #define ECB2_ESCA    0x04
+#define ECB2_ZPCI_LSI    0x02

      __u8    ecb2;                   /* 0x0062 */

arch/s390/include/asm/kvm_host.h +2 -0
+#define ECB3_AISI    0x20
+#define ECB3_AISII    0x10

  #define ECB3_DEA 0x08
  #define ECB3_AES 0x04
  #define ECB3_RI  0x01

arch/s390/include/asm/kvm_host.h +1 -0
@@ -938,6 +941,7 @@ struct kvm_arch{
       int use_cmma;
       int use_pfmfi;
       int use_skf;
+    int use_zpci_interp;
       int user_cpu_state_ctrl;
       int user_sigp;
       int user_stsi;
arch/s390/kvm/kvm-s390.c +2 -0
index ab8b56deed11..b6c32fc3b272 100644
@@ -1029,6 +1029,44 @@ static int kvm_s390_vm_set_crypto(struct kvm 
 *kvm, struct kvm_device_attr *attr)
       return 0;
   }
+static void kvm_s390_vcpu_pci_setup(struct kvm_vcpu *vcpu)
+{
Quoted 9 lines

arch/s390/kvm/kvm-s390.c static int kvm_s390_vm_set_crypto(struct kvm *kvm, struct kvm_device_attr *attr) · line 1039

Quoted 6 lines
arch/s390/kvm/kvm-s390.c +6 -0
+    /* Only set the ECB bits after guest requests zPCI interpretation */
+    if (!vcpu->kvm->arch.use_zpci_interp)
+        return;
+
+    vcpu->arch.sie_block->ecb2 |= ECB2_ZPCI_LSI;
+    vcpu->arch.sie_block->ecb3 |= ECB3_AISII + ECB3_AISI;

As far as I understood, the interpretation is only possible if a gisa designation is associated with the PCI function via CLP enable.

This is true. Once ECB is enabled, you must have either a SHM bit on for emulated device support or SHM bits off + a GISA designation registered for interpretation. Otherwise, PCI instructions will fail.

Quoted 1 line

Why do we setup the SIE ECB only when the guest requests for interpretation and not systematically in vcpu_setup?

Once the ECB is enabled for a guest, emulated device FHs must have a SHM bit in order to continue working properly (so do passthrough devices that don't setup interpretation). This was not a requirement before this series -- simply having the ECB bit off would ensure intercepts for all devices regardless of SHM bit settings, so by doing an opt-in once the guest indicates it will be doing interpretation we can preserve backwards-compatibility with an initial mode where SHM bits are not necessarily required. However once userspace indicates it understands interpretation, we can assume it is will also use SHM bits properly.

Quoted 3 lines

If ECB2_ZPCI_LSI, ECB3_AISII or ECB3_AISI have an effect when the gisa designation is not specified shouldn't we have a way to clear these bits?

I'm not sure that's necessary -- The idea here was for the userspace to indicate 1) that it knows how to setup for interpreted devices and 2) that it has a guest that wants to use at least 1 interpreted device.
Once we know that userspace understands how to manage interpreted devices (implied by its use of these new vfio feature ioctls) I think it should be OK to leave these bits on and expect userspace to always do the appropriate steps (SHM bits for emulated devices / forced intercept passthrough devices, GISA designation for interpreted devices).

Acked-by
#49 lore raw

Re: [PATCH v2 17/30] KVM: s390: mechanism to enable guest zPCI Interpretation

Quoted 62 lines · Matthew Rosato
Quoted 51 lines · Pierre Morel
Quoted 48 lines · Matthew Rosato

The guest must have access to certain facilities in order to allow interpretive execution of zPCI instructions and adapter event notifications.  However, there are some cases where a guest might disable interpretation -- provide a mechanism via which we can defer enabling the associated zPCI interpretation facilities until the guest indicates it wishes to use them.

---
  arch/s390/include/asm/kvm_host.h |  4 ++++ arch/s390/kvm/kvm-s390.c         | 40 ++++++++++++++++++++++++++++++++ arch/s390/kvm/kvm-s390.h         | 10 ++++++++

  3 files changed, 54 insertions(+)
arch/s390/include/asm/kvm_host.h +1 -0
index 3f147b8d050b..38982c1de413 100644
@@ -252,7 +252,10 @@ struct kvm_s390_sie_block {
   #define ECB2_IEP    0x20
   #define ECB2_PFMFI    0x08
   #define ECB2_ESCA    0x04
+#define ECB2_ZPCI_LSI    0x02

      __u8    ecb2;                   /* 0x0062 */

arch/s390/include/asm/kvm_host.h +2 -0
+#define ECB3_AISI    0x20
+#define ECB3_AISII    0x10

  #define ECB3_DEA 0x08
  #define ECB3_AES 0x04
  #define ECB3_RI  0x01

arch/s390/include/asm/kvm_host.h +1 -0
@@ -938,6 +941,7 @@ struct kvm_arch{
       int use_cmma;
       int use_pfmfi;
       int use_skf;
+    int use_zpci_interp;
       int user_cpu_state_ctrl;
       int user_sigp;
       int user_stsi;
arch/s390/kvm/kvm-s390.c +3 -0
index ab8b56deed11..b6c32fc3b272 100644
@@ -1029,6 +1029,44 @@ static int kvm_s390_vm_set_crypto(struct kvm 
 *kvm, struct kvm_device_attr *attr)
       return 0;
   }
+static void kvm_s390_vcpu_pci_setup(struct kvm_vcpu *vcpu)
+{
+    /* Only set the ECB bits after guest requests zPCI 
Quoted 9 lines
Quoted 6 lines

interpretation */
+    if (!vcpu->kvm->arch.use_zpci_interp)
+        return;
+
+    vcpu->arch.sie_block->ecb2 |= ECB2_ZPCI_LSI; +    vcpu->arch.sie_block->ecb3 |= ECB3_AISII + ECB3_AISI;

As far as I understood, the interpretation is only possible if a gisa designation is associated with the PCI function via CLP enable.

Quoted 1 line

This is true.  Once ECB is enabled, you must have either a SHM bit on for emulated device support or SHM bits off + a GISA designation registered for interpretation.  Otherwise, PCI instructions will fail.

AFAIU the PCI instruction should not fail but trigger an interception if the GISA designation field of the CLP enable.

However, what you do is not false.
So I think we better keep what you propose.

Quoted 4 lines
Quoted 2 lines

Why do we setup the SIE ECB only when the guest requests for interpretation and not systematically in vcpu_setup?

Once the ECB is enabled for a guest, emulated device FHs must have a SHM bit in order to continue working properly (so do passthrough devices that don't setup interpretation).  This was not a requirement before this series -- simply having the ECB bit off would ensure intercepts for all devices regardless of SHM bit settings, so by doing an opt-in once the guest indicates it will be doing interpretation we can preserve backwards-compatibility with an initial mode where SHM bits are not necessarily required.  However once userspace indicates it understands interpretation, we can assume it is will also use SHM bits properly.

If not setting GD in CLP enable triggers interception on later PCI instructions, preparing all early would allow to chose between interpretation or interception on a function basis during the CLP set PCI function with the enable command and make the initialization simpler.

However, what you propose is tested and works so we can have this discussion later for enhancement, if it is really one.

Quoted 7 lines
Quoted 4 lines

If ECB2_ZPCI_LSI, ECB3_AISII or ECB3_AISI have an effect when the gisa designation is not specified shouldn't we have a way to clear these bits?

I'm not sure that's necessary -- The idea here was for the userspace to indicate 1) that it knows how to setup for interpreted devices and 2) that it has a guest that wants to use at least 1 interpreted device.
Once we know that userspace understands how to manage interpreted devices (implied by its use of these new vfio feature ioctls) I think it should be OK to leave these bits on and expect userspace to always do the appropriate steps (SHM bits for emulated devices / forced intercept passthrough devices, GISA designation for interpreted devices).

Seems reasonable.

Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#50 lore raw

PATCH v2 18/30 KVM: s390: pci: provide routines for enabling/disabling interpretation

In reply to Matthew Rosato

These routines will be wired into the vfio_pci_zdev ioctl handlers to respond to requests to enable / disable a device for zPCI Load/Store interpretation.

The first time such a request is received, enable the necessary facilities for the guest.

---
 arch/s390/include/asm/kvm_pci.h |  4 ++
 arch/s390/kvm/pci.c             | 99 +++++++++++++++++++++++++++++++++
 arch/s390/pci/pci.c             |  3 +
 3 files changed, 106 insertions(+)
arch/s390/include/asm/kvm_pci.h +4 -0
index aafee2976929..072401aa7922 100644
@@ -26,4 +26,8 @@ int kvm_s390_pci_dev_open(struct zpci_dev *zdev);
 void kvm_s390_pci_dev_release(struct zpci_dev *zdev);
 void kvm_s390_pci_attach_kvm(struct zpci_dev *zdev, struct kvm *kvm);
 
+int kvm_s390_pci_interp_probe(struct zpci_dev *zdev);
+int kvm_s390_pci_interp_enable(struct zpci_dev *zdev);
+int kvm_s390_pci_interp_disable(struct zpci_dev *zdev);
+
 #endif /* ASM_KVM_PCI_H */
arch/s390/kvm/pci.c +99 -0

112 lines not on this page · raw message

arch/s390/pci/pci.c +3 -0
index 2a19becbc14c..58673f633869 100644
@@ -147,6 +147,7 @@ int zpci_register_ioat(struct zpci_dev *zdev, u8 dmaas,
 		zpci_dbg(3, "reg ioat fid:%x, cc:%d, status:%d\n", zdev->fid, cc, status);
 	return cc;
 }
+EXPORT_SYMBOL_GPL(zpci_register_ioat);
 
 /* Modify PCI: Unregister I/O address translation parameters */
 int zpci_unregister_ioat(struct zpci_dev *zdev, u8 dmaas)
@@ -727,6 +728,7 @@ int zpci_enable_device(struct zpci_dev *zdev)
 		zpci_update_fh(zdev, fh);
 	return rc;
 }
+EXPORT_SYMBOL_GPL(zpci_enable_device);
 
 int zpci_disable_device(struct zpci_dev *zdev)
 {
@@ -750,6 +752,7 @@ int zpci_disable_device(struct zpci_dev *zdev)
 	}
 	return rc;
 }
+EXPORT_SYMBOL_GPL(zpci_disable_device);
 
 /**
  * zpci_hot_reset_device - perform a reset of the given zPCI function
Signature · 1 line
2.27.0
#51 lore raw

Re: [PATCH v2 18/30] KVM: s390: pci: provide routines for enabling/disabling interpretation

Quoted 42 lines · Matthew Rosato

These routines will be wired into the vfio_pci_zdev ioctl handlers to respond to requests to enable / disable a device for zPCI Load/Store interpretation.

The first time such a request is received, enable the necessary facilities for the guest.

---
  arch/s390/include/asm/kvm_pci.h |  4 ++
  arch/s390/kvm/pci.c             | 99 +++++++++++++++++++++++++++++++++
  arch/s390/pci/pci.c             |  3 +
  3 files changed, 106 insertions(+)
arch/s390/include/asm/kvm_pci.h +4 -0
index aafee2976929..072401aa7922 100644
@@ -26,4 +26,8 @@ int kvm_s390_pci_dev_open(struct zpci_dev *zdev);
  void kvm_s390_pci_dev_release(struct zpci_dev *zdev);
  void kvm_s390_pci_attach_kvm(struct zpci_dev *zdev, struct kvm *kvm);
 
+int kvm_s390_pci_interp_probe(struct zpci_dev *zdev);
+int kvm_s390_pci_interp_enable(struct zpci_dev *zdev);
+int kvm_s390_pci_interp_disable(struct zpci_dev *zdev);
+
  #endif /* ASM_KVM_PCI_H */
arch/s390/kvm/pci.c +2 -0
index dae853da6df1..122d0992b521 100644
@@ -12,7 +12,9 @@
  #include <asm/kvm_pci.h>
  #include <asm/pci.h>
  #include <asm/pci_insn.h>
+#include <asm/sclp.h>
  #include "pci.h"
+#include "kvm-s390.h"
 
  struct zpci_aift *aift;
 

arch/s390/kvm/pci.c int kvm_s390_pci_aen_init(u8 nisc) · line 151

Quoted 7 lines
arch/s390/kvm/pci.c +4 -0
@@ -143,6 +145,103 @@ int kvm_s390_pci_aen_init(u8 nisc)
  	return rc;
  }
 
+int kvm_s390_pci_interp_probe(struct zpci_dev *zdev)
+{
+	/* Must have appropriate hardware facilities */
+	if (!(sclp.has_zpci_lsi && test_facility(69)))

Should'nt we also test the other facilities we need for the interpretation like ARNI, AISII, ASI and GISA ?

Or are we sure they are always there when ZPCI load/store interpretation is available?

Quoted 21 lines
arch/s390/kvm/pci.c +21 -0
+		return -EINVAL;
+
+	/* Must have a KVM association registered */
+	if (!zdev->kzdev || !zdev->kzdev->kvm)
+		return -EINVAL;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(kvm_s390_pci_interp_probe);
+
+int kvm_s390_pci_interp_enable(struct zpci_dev *zdev)
+{
+	u32 gd;
+	int rc;
+
+	if (!zdev->kzdev || !zdev->kzdev->kvm)
+		return -EINVAL;
+
+	/*
+	 * If this is the first request to use an interpreted device, make the
+	 * necessary vcpu changes

arch/s390/kvm/pci.c int kvm_s390_pci_aen_init(u8 nisc) · line 182

Quoted 10 lines
arch/s390/kvm/pci.c +10 -0
+	 */
+	if (!zdev->kzdev->kvm->arch.use_zpci_interp)
+		kvm_s390_vcpu_pci_enable_interp(zdev->kzdev->kvm);
+
+	/*
+	 * In the event of a system reset in userspace, the GISA designation
+	 * may still be assigned because the device is still enabled.
+	 * Verify it's the same guest before proceeding.
+	 */
+	gd = (u32)(u64)&zdev->kzdev->kvm->arch.sie_page2->gisa;

should use the virt_to_phys transformation ?

Quoted 94 lines
arch/s390/kvm/pci.c +62 -0
+	if (zdev->gd != 0 && zdev->gd != gd)
+		return -EPERM;
+
+	if (zdev_enabled(zdev)) {
+		zdev->gd = 0;
+		rc = zpci_disable_device(zdev);
+		if (rc)
+			return rc;
+	}
+
+	/*
+	 * Store information about the identity of the kvm guest allowed to
+	 * access this device via interpretation to be used by host CLP
+	 */
+	zdev->gd = gd;
+
+	rc = zpci_enable_device(zdev);
+	if (rc)
+		goto err;
+
+	/* Re-register the IOMMU that was already created */
+	rc = zpci_register_ioat(zdev, 0, zdev->start_dma, zdev->end_dma,
+				virt_to_phys(zdev->dma_table));
+	if (rc)
+		goto err;
+
+	return rc;
+
+err:
+	zdev->gd = 0;
+	return rc;
+}
+EXPORT_SYMBOL_GPL(kvm_s390_pci_interp_enable);
+
+int kvm_s390_pci_interp_disable(struct zpci_dev *zdev)
+{
+	int rc;
+
+	if (zdev->gd == 0)
+		return -EINVAL;
+
+	/* Remove the host CLP guest designation */
+	zdev->gd = 0;
+
+	if (zdev_enabled(zdev)) {
+		rc = zpci_disable_device(zdev);
+		if (rc)
+			return rc;
+	}
+
+	rc = zpci_enable_device(zdev);
+	if (rc)
+		return rc;
+
+	/* Re-register the IOMMU that was already created */
+	rc = zpci_register_ioat(zdev, 0, zdev->start_dma, zdev->end_dma,
+				virt_to_phys(zdev->dma_table));
+
+	return rc;
+}
+EXPORT_SYMBOL_GPL(kvm_s390_pci_interp_disable);
+
  int kvm_s390_pci_dev_open(struct zpci_dev *zdev)
  {
  	struct kvm_zdev *kzdev;
arch/s390/pci/pci.c +3 -0
index 2a19becbc14c..58673f633869 100644
@@ -147,6 +147,7 @@ int zpci_register_ioat(struct zpci_dev *zdev, u8 dmaas,
  		zpci_dbg(3, "reg ioat fid:%x, cc:%d, status:%d\n", zdev->fid, cc, status);
  	return cc;
  }
+EXPORT_SYMBOL_GPL(zpci_register_ioat);
 
  /* Modify PCI: Unregister I/O address translation parameters */
  int zpci_unregister_ioat(struct zpci_dev *zdev, u8 dmaas)
@@ -727,6 +728,7 @@ int zpci_enable_device(struct zpci_dev *zdev)
  		zpci_update_fh(zdev, fh);
  	return rc;
  }
+EXPORT_SYMBOL_GPL(zpci_enable_device);
 
  int zpci_disable_device(struct zpci_dev *zdev)
  {
@@ -750,6 +752,7 @@ int zpci_disable_device(struct zpci_dev *zdev)
  	}
  	return rc;
  }
+EXPORT_SYMBOL_GPL(zpci_disable_device);
 
  /**
   * zpci_hot_reset_device - perform a reset of the given zPCI function
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#52 lore raw

Re: [PATCH v2 18/30] KVM: s390: pci: provide routines for enabling/disabling interpretation

Quoted 51 lines · Pierre Morel
Quoted 39 lines · Matthew Rosato

These routines will be wired into the vfio_pci_zdev ioctl handlers to respond to requests to enable / disable a device for zPCI Load/Store interpretation.

The first time such a request is received, enable the necessary facilities for the guest.

---
  arch/s390/include/asm/kvm_pci.h |  4 ++ arch/s390/kvm/pci.c             | 99 +++++++++++++++++++++++++++++++++ arch/s390/pci/pci.c             |  3 +

  3 files changed, 106 insertions(+)
arch/s390/include/asm/kvm_pci.h +4 -0
index aafee2976929..072401aa7922 100644
@@ -26,4 +26,8 @@ int kvm_s390_pci_dev_open(struct zpci_dev *zdev);
   void kvm_s390_pci_dev_release(struct zpci_dev *zdev);
   void kvm_s390_pci_attach_kvm(struct zpci_dev *zdev, struct kvm *kvm);
+int kvm_s390_pci_interp_probe(struct zpci_dev *zdev);
+int kvm_s390_pci_interp_enable(struct zpci_dev *zdev);
+int kvm_s390_pci_interp_disable(struct zpci_dev *zdev);
+
   #endif /* ASM_KVM_PCI_H */
arch/s390/kvm/pci.c +2 -0
index dae853da6df1..122d0992b521 100644
@@ -12,7 +12,9 @@
   #include <asm/kvm_pci.h>
   #include <asm/pci.h>
   #include <asm/pci_insn.h>
+#include <asm/sclp.h>
   #include "pci.h"
+#include "kvm-s390.h"
   struct zpci_aift *aift;

arch/s390/kvm/pci.c int kvm_s390_pci_aen_init(u8 nisc) · line 150

Quoted 6 lines
arch/s390/kvm/pci.c +4 -0
@@ -143,6 +145,103 @@ int kvm_s390_pci_aen_init(u8 nisc)
       return rc;
   }
+int kvm_s390_pci_interp_probe(struct zpci_dev *zdev)
+{
+    /* Must have appropriate hardware facilities */
+    if (!(sclp.has_zpci_lsi && test_facility(69)))

Should'nt we also test the other facilities we need for the interpretation like ARNI, AISII, ASI and GISA ?

Quoted 1 line

Or are we sure they are always there when ZPCI load/store interpretation is available?

I think some of these are implicit based on others but I think you're right that we should be testing for more than this to be safe. I think additionally test for AENI, AISII, AISI -- basically we should match what we test for in patch 17.

Quoted 28 lines
Quoted 28 lines
arch/s390/kvm/pci.c +20 -0
+        return -EINVAL;
+
+    /* Must have a KVM association registered */
+    if (!zdev->kzdev || !zdev->kzdev->kvm)
+        return -EINVAL;
+
+    return 0;
+}
+EXPORT_SYMBOL_GPL(kvm_s390_pci_interp_probe);
+
+int kvm_s390_pci_interp_enable(struct zpci_dev *zdev)
+{
+    u32 gd;
+    int rc;
+
+    if (!zdev->kzdev || !zdev->kzdev->kvm)
+        return -EINVAL;
+
+    /*
+     * If this is the first request to use an interpreted device, 

make the
+     * necessary vcpu changes
+     */
+    if (!zdev->kzdev->kvm->arch.use_zpci_interp) +        kvm_s390_vcpu_pci_enable_interp(zdev->kzdev->kvm); +

Quoted 8 lines
Quoted 6 lines

+    /*
+     * In the event of a system reset in userspace, the GISA designation +     * may still be assigned because the device is still enabled. +     * Verify it's the same guest before proceeding.
+     */
+    gd = (u32)(u64)&zdev->kzdev->kvm->arch.sie_page2->gisa;

should use the virt_to_phys transformation ?

Yes

author maintainer
#53 lore raw

PATCH v2 19/30 KVM: s390: pci: provide routines for enabling/disabling interrupt forwarding

In reply to Matthew Rosato

These routines will be wired into the vfio_pci_zdev ioctl handlers to respond to requests to enable / disable a device for Adapter Event Notifications / Adapter Interuption Forwarding.

---
 arch/s390/include/asm/kvm_pci.h |   7 ++
 arch/s390/kvm/pci.c             | 203 ++++++++++++++++++++++++++++++++
 arch/s390/pci/pci_insn.c        |   1 +
 3 files changed, 211 insertions(+)
arch/s390/include/asm/kvm_pci.h +7 -0
index 072401aa7922..01fe14fffd7a 100644
@@ -16,16 +16,23 @@
 #include <linux/kvm_host.h>
 #include <linux/kvm.h>
 #include <linux/pci.h>
+#include <asm/pci_insn.h>
 
 struct kvm_zdev {
 	struct zpci_dev *zdev;
 	struct kvm *kvm;
+	struct zpci_fib fib;
 };
 
 int kvm_s390_pci_dev_open(struct zpci_dev *zdev);
 void kvm_s390_pci_dev_release(struct zpci_dev *zdev);
 void kvm_s390_pci_attach_kvm(struct zpci_dev *zdev, struct kvm *kvm);
 
+int kvm_s390_pci_aif_probe(struct zpci_dev *zdev);
+int kvm_s390_pci_aif_enable(struct zpci_dev *zdev, struct zpci_fib *fib,
+			    bool assist);
+int kvm_s390_pci_aif_disable(struct zpci_dev *zdev);
+
 int kvm_s390_pci_interp_probe(struct zpci_dev *zdev);
 int kvm_s390_pci_interp_enable(struct zpci_dev *zdev);
 int kvm_s390_pci_interp_disable(struct zpci_dev *zdev);
arch/s390/kvm/pci.c +203 -0

221 lines not on this page · raw message

arch/s390/pci/pci_insn.c +1 -0
index ca6399d52767..f7d0e29bbf0b 100644
@@ -59,6 +59,7 @@ u8 zpci_mod_fc(u64 req, struct zpci_fib *fib, u8 *status)
 
 	return cc;
 }
+EXPORT_SYMBOL_GPL(zpci_mod_fc);
 
 /* Refresh PCI Translations */
 static inline u8 __rpcit(u64 fn, u64 addr, u64 range, u8 *status)
Signature · 1 line
2.27.0
#54 lore raw

Re: [PATCH v2 19/30] KVM: s390: pci: provide routines for enabling/disabling interrupt forwarding

Quoted 52 lines · Matthew Rosato

These routines will be wired into the vfio_pci_zdev ioctl handlers to respond to requests to enable / disable a device for Adapter Event Notifications / Adapter Interuption Forwarding.

---
  arch/s390/include/asm/kvm_pci.h |   7 ++
  arch/s390/kvm/pci.c             | 203 ++++++++++++++++++++++++++++++++
  arch/s390/pci/pci_insn.c        |   1 +
  3 files changed, 211 insertions(+)
arch/s390/include/asm/kvm_pci.h +7 -0
index 072401aa7922..01fe14fffd7a 100644
@@ -16,16 +16,23 @@
  #include <linux/kvm_host.h>
  #include <linux/kvm.h>
  #include <linux/pci.h>
+#include <asm/pci_insn.h>
 
  struct kvm_zdev {
  	struct zpci_dev *zdev;
  	struct kvm *kvm;
+	struct zpci_fib fib;
  };
 
  int kvm_s390_pci_dev_open(struct zpci_dev *zdev);
  void kvm_s390_pci_dev_release(struct zpci_dev *zdev);
  void kvm_s390_pci_attach_kvm(struct zpci_dev *zdev, struct kvm *kvm);
 
+int kvm_s390_pci_aif_probe(struct zpci_dev *zdev);
+int kvm_s390_pci_aif_enable(struct zpci_dev *zdev, struct zpci_fib *fib,
+			    bool assist);
+int kvm_s390_pci_aif_disable(struct zpci_dev *zdev);
+
  int kvm_s390_pci_interp_probe(struct zpci_dev *zdev);
  int kvm_s390_pci_interp_enable(struct zpci_dev *zdev);
  int kvm_s390_pci_interp_disable(struct zpci_dev *zdev);
arch/s390/kvm/pci.c +1 -0
index 122d0992b521..7ed9abc476b6 100644
@@ -12,6 +12,7 @@
  #include <asm/kvm_pci.h>
  #include <asm/pci.h>
  #include <asm/pci_insn.h>
+#include <asm/pci_io.h>
  #include <asm/sclp.h>
  #include "pci.h"
  #include "kvm-s390.h"

arch/s390/kvm/pci.c int kvm_s390_pci_aen_init(u8 nisc) · line 153

Quoted 8 lines
arch/s390/kvm/pci.c +5 -0
@@ -145,6 +146,204 @@ int kvm_s390_pci_aen_init(u8 nisc)
  	return rc;
  }
 
+/* Modify PCI: Register floating adapter interruption forwarding */
+static int kvm_zpci_set_airq(struct zpci_dev *zdev)
+{
+	u64 req = ZPCI_CREATE_REQ(zdev->fh, 0, ZPCI_MOD_FC_REG_INT);
+	struct zpci_fib fib = {0};

I prefer {} instead of {0} even it does the same it looks wrong to me.

Quoted 9 lines
arch/s390/kvm/pci.c +9 -0
+	u8 status;
+
+	fib.fmt0.isc = zdev->kzdev->fib.fmt0.isc;
+	fib.fmt0.sum = 1;       /* enable summary notifications */
+	fib.fmt0.noi = airq_iv_end(zdev->aibv);
+	fib.fmt0.aibv = virt_to_phys(zdev->aibv->vector);
+	fib.fmt0.aibvo = 0;
+	fib.fmt0.aisb = virt_to_phys(aift->sbv->vector + (zdev->aisb / 64) * 8);
+	fib.fmt0.aisbo = zdev->aisb & 63;

arch/s390/kvm/pci.c int kvm_s390_pci_aen_init(u8 nisc) · line 172

Quoted 10 lines
arch/s390/kvm/pci.c +10 -0
+	fib.gd = zdev->gd;
+
+	return zpci_mod_fc(req, &fib, &status) ? -EIO : 0;
+}
+
+/* Modify PCI: Unregister floating adapter interruption forwarding */
+static int kvm_zpci_clear_airq(struct zpci_dev *zdev)
+{
+	u64 req = ZPCI_CREATE_REQ(zdev->fh, 0, ZPCI_MOD_FC_DEREG_INT);
+	struct zpci_fib fib = {0};

same here

Quoted 65 lines
arch/s390/kvm/pci.c +65 -0
+	u8 cc, status;
+
+	fib.gd = zdev->gd;
+
+	cc = zpci_mod_fc(req, &fib, &status);
+	if (cc == 3 || (cc == 1 && status == 24))
+		/* Function already gone or IRQs already deregistered. */
+		cc = 0;
+
+	return cc ? -EIO : 0;
+}
+
+int kvm_s390_pci_aif_probe(struct zpci_dev *zdev)
+{
+	/* Must have appropriate hardware facilities */
+	if (!(sclp.has_aeni && test_facility(71)))
+		return -EINVAL;
+
+	/* Must have a KVM association registered */
+	if (!zdev->kzdev || !zdev->kzdev->kvm)
+		return -EINVAL;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(kvm_s390_pci_aif_probe);
+
+int kvm_s390_pci_aif_enable(struct zpci_dev *zdev, struct zpci_fib *fib,
+			    bool assist)
+{
+	struct page *aibv_page, *aisb_page = NULL;
+	unsigned int msi_vecs, idx;
+	struct zpci_gaite *gaite;
+	unsigned long bit;
+	struct kvm *kvm;
+	phys_addr_t gaddr;
+	int rc = 0;
+
+	/*
+	 * Interrupt forwarding is only applicable if the device is already
+	 * enabled for interpretation
+	 */
+	if (zdev->gd == 0)
+		return -EINVAL;
+
+	kvm = zdev->kzdev->kvm;
+	msi_vecs = min_t(unsigned int, fib->fmt0.noi, zdev->max_msi);
+
+	/* Replace AIBV address */
+	idx = srcu_read_lock(&kvm->srcu);
+	aibv_page = gfn_to_page(kvm, gpa_to_gfn((gpa_t)fib->fmt0.aibv));
+	srcu_read_unlock(&kvm->srcu, idx);
+	if (is_error_page(aibv_page)) {
+		rc = -EIO;
+		goto out;
+	}
+	gaddr = page_to_phys(aibv_page) + (fib->fmt0.aibv & ~PAGE_MASK);
+	fib->fmt0.aibv = gaddr;
+
+	/* Pin the guest AISB if one was specified */
+	if (fib->fmt0.sum == 1) {
+		idx = srcu_read_lock(&kvm->srcu);
+		aisb_page = gfn_to_page(kvm, gpa_to_gfn((gpa_t)fib->fmt0.aisb));
+		srcu_read_unlock(&kvm->srcu, idx);
+		if (is_error_page(aisb_page)) {
+			rc = -EIO;

arch/s390/kvm/pci.c int kvm_s390_pci_aen_init(u8 nisc) · line 247

Quoted 10 lines
arch/s390/kvm/pci.c +10 -0
+			goto unpin1;
+		}
+	}
+
+	/* AISB must be allocated before we can fill in GAITE */
+	mutex_lock(&aift->lock);
+	bit = airq_iv_alloc_bit(aift->sbv);
+	if (bit == -1UL)
+		goto unpin2;
+	zdev->aisb = bit;

aisb here is the aisb offset right?
Then may be add a comment as in gait and fmt0 aisb is an address.

arch/s390/kvm/pci.c int kvm_s390_pci_aen_init(u8 nisc) · line 251

Quoted 4 lines
arch/s390/kvm/pci.c +4 -0
+	zdev->aibv = airq_iv_create(msi_vecs, AIRQ_IV_DATA |
+					      AIRQ_IV_BITLOCK |
+					      AIRQ_IV_GUESTVEC,
+				    (unsigned long *)fib->fmt0.aibv);

phys_to_virt ?

arch/s390/kvm/pci.c int kvm_s390_pci_aen_init(u8 nisc) · line 259

Quoted 8 lines
arch/s390/kvm/pci.c +8 -0
+
+	spin_lock_irq(&aift->gait_lock);
+	gaite = (struct zpci_gaite *)aift->gait + (zdev->aisb *
+						   sizeof(struct zpci_gaite));
+
+	/* If assist not requested, host will get all alerts */
+	if (assist)
+		gaite->gisa = (u32)(u64)&kvm->arch.sie_page2->gisa;

virt_to_phys ?

Quoted 40 lines
arch/s390/kvm/pci.c +40 -0
+	else
+		gaite->gisa = 0;
+
+	gaite->gisc = fib->fmt0.isc;
+	gaite->count++;
+	gaite->aisbo = fib->fmt0.aisbo;
+	gaite->aisb = virt_to_phys(page_address(aisb_page) + (fib->fmt0.aisb &
+							      ~PAGE_MASK));
+	aift->kzdev[zdev->aisb] = zdev->kzdev;
+	spin_unlock_irq(&aift->gait_lock);
+
+	/* Update guest FIB for re-issue */
+	fib->fmt0.aisbo = zdev->aisb & 63;
+	fib->fmt0.aisb = virt_to_phys(aift->sbv->vector + (zdev->aisb / 64) * 8);
+	fib->fmt0.isc = kvm_s390_gisc_register(kvm, gaite->gisc);
+
+	/* Save some guest fib values in the host for later use */
+	zdev->kzdev->fib.fmt0.isc = fib->fmt0.isc;
+	zdev->kzdev->fib.fmt0.aibv = fib->fmt0.aibv;
+	mutex_unlock(&aift->lock);
+
+	/* Issue the clp to setup the irq now */
+	rc = kvm_zpci_set_airq(zdev);
+	return rc;
+
+unpin2:
+	mutex_unlock(&aift->lock);
+	if (fib->fmt0.sum == 1) {
+		gaddr = page_to_phys(aisb_page);
+		kvm_release_pfn_dirty(gaddr >> PAGE_SHIFT);
+	}
+unpin1:
+	kvm_release_pfn_dirty(fib->fmt0.aibv >> PAGE_SHIFT);
+out:
+	return rc;
+}
+EXPORT_SYMBOL_GPL(kvm_s390_pci_aif_enable);
+
+int kvm_s390_pci_aif_disable(struct zpci_dev *zdev)
+{

arch/s390/kvm/pci.c int kvm_s390_pci_aen_init(u8 nisc) · line 309

Quoted 10 lines
arch/s390/kvm/pci.c +10 -0
+	struct kvm_zdev *kzdev = zdev->kzdev;
+	struct zpci_gaite *gaite;
+	int rc;
+	u8 isc;
+
+	if (zdev->gd == 0)
+		return -EINVAL;
+
+	/* Even if the clear fails due to an error, clear the GAITE */
+	rc = kvm_zpci_clear_airq(zdev);

Having a look at kvm_zpci_clear_airq() the only possible error seems to be when an error recovery is in progress.
The error returned for a wrong FH, function does not exist anymore, or if the interrupt vectors are already deregistered by the instruction are returned as success by the function.

How can we be sure that we have no conflict with a recovery in progress? Shouldn't we in this case let the recovery process handle the function and stop here?

Doesn't the aif lock mutex placed after and not before the clear_irq open a door for race condition with the recovery?

Quoted 18 lines
arch/s390/kvm/pci.c +18 -0
+
+	mutex_lock(&aift->lock);
+	if (zdev->kzdev->fib.fmt0.aibv == 0)
+		goto out;
+	spin_lock_irq(&aift->gait_lock);
+	gaite = (struct zpci_gaite *)aift->gait + (zdev->aisb *
+						   sizeof(struct zpci_gaite));
+	isc = gaite->gisc;
+	gaite->count--;
+	if (gaite->count == 0) {
+		/* Release guest AIBV and AISB */
+		kvm_release_pfn_dirty(kzdev->fib.fmt0.aibv >> PAGE_SHIFT);
+		if (gaite->aisb != 0)
+			kvm_release_pfn_dirty(gaite->aisb >> PAGE_SHIFT);
+		/* Clear the GAIT entry */
+		gaite->aisb = 0;
+		gaite->gisc = 0;
+		gaite->aisbo = 0;

arch/s390/kvm/pci.c int kvm_s390_pci_aen_init(u8 nisc) · line 337

Quoted 10 lines
arch/s390/kvm/pci.c +10 -0
+		gaite->gisa = 0;
+		aift->kzdev[zdev->aisb] = 0;
+		/* Clear zdev info */
+		airq_iv_free_bit(aift->sbv, zdev->aisb);
+		airq_iv_release(zdev->aibv);
+		zdev->aisb = 0;
+		zdev->aibv = NULL;
+	}
+	spin_unlock_irq(&aift->gait_lock);
+	kvm_s390_gisc_unregister(kzdev->kvm, isc);

Don't we need to check the return value?
And maybe to report it to the caller?

Quoted 36 lines
arch/s390/kvm/pci.c +13 -0
+	kzdev->fib.fmt0.isc = 0;
+	kzdev->fib.fmt0.aibv = 0;
+out:
+	mutex_unlock(&aift->lock);
+
+	return rc;
+}
+EXPORT_SYMBOL_GPL(kvm_s390_pci_aif_disable);
+
  int kvm_s390_pci_interp_probe(struct zpci_dev *zdev)
  {
  	/* Must have appropriate hardware facilities */
@@ -221,6 +420,10 @@ int kvm_s390_pci_interp_disable(struct zpci_dev *zdev)
  	if (zdev->gd == 0)
  		return -EINVAL;
 
+	/* Forwarding must be turned off before interpretation */
+	if (zdev->kzdev->fib.fmt0.aibv != 0)
+		kvm_s390_pci_aif_disable(zdev);
+
  	/* Remove the host CLP guest designation */
  	zdev->gd = 0;
 
arch/s390/pci/pci_insn.c +1 -0
index ca6399d52767..f7d0e29bbf0b 100644
@@ -59,6 +59,7 @@ u8 zpci_mod_fc(u64 req, struct zpci_fib *fib, u8 *status)
 
  	return cc;
  }
+EXPORT_SYMBOL_GPL(zpci_mod_fc);
 
  /* Refresh PCI Translations */
  static inline u8 __rpcit(u64 fn, u64 addr, u64 range, u8 *status)
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#55 lore raw

Re: [PATCH v2 19/30] KVM: s390: pci: provide routines for enabling/disabling interrupt forwarding

Quoted 3 lines · Pierre Morel

On 1/14/22 21:31, Matthew Rosato wrote:

...

Quoted 8 lines

arch/s390/kvm/pci.c int kvm_s390_pci_aen_init(u8 nisc) · line 153

Quoted 5 lines
arch/s390/kvm/pci.c +5 -0
+/* Modify PCI: Register floating adapter interruption forwarding */
+static int kvm_zpci_set_airq(struct zpci_dev *zdev)
+{
+    u64 req = ZPCI_CREATE_REQ(zdev->fh, 0, ZPCI_MOD_FC_REG_INT);
+    struct zpci_fib fib = {0};

I prefer {} instead of {0} even it does the same it looks wrong to me.

OK

...

Quoted 43 lines
Quoted 43 lines
arch/s390/kvm/pci.c +43 -0
+int kvm_s390_pci_aif_enable(struct zpci_dev *zdev, struct zpci_fib *fib,
+                bool assist)
+{
+    struct page *aibv_page, *aisb_page = NULL;
+    unsigned int msi_vecs, idx;
+    struct zpci_gaite *gaite;
+    unsigned long bit;
+    struct kvm *kvm;
+    phys_addr_t gaddr;
+    int rc = 0;
+
+    /*
+     * Interrupt forwarding is only applicable if the device is already
+     * enabled for interpretation
+     */
+    if (zdev->gd == 0)
+        return -EINVAL;
+
+    kvm = zdev->kzdev->kvm;
+    msi_vecs = min_t(unsigned int, fib->fmt0.noi, zdev->max_msi);
+
+    /* Replace AIBV address */
+    idx = srcu_read_lock(&kvm->srcu);
+    aibv_page = gfn_to_page(kvm, gpa_to_gfn((gpa_t)fib->fmt0.aibv));
+    srcu_read_unlock(&kvm->srcu, idx);
+    if (is_error_page(aibv_page)) {
+        rc = -EIO;
+        goto out;
+    }
+    gaddr = page_to_phys(aibv_page) + (fib->fmt0.aibv & ~PAGE_MASK);
+    fib->fmt0.aibv = gaddr;
+
+    /* Pin the guest AISB if one was specified */
+    if (fib->fmt0.sum == 1) {
+        idx = srcu_read_lock(&kvm->srcu);
+        aisb_page = gfn_to_page(kvm, gpa_to_gfn((gpa_t)fib->fmt0.aisb));
+        srcu_read_unlock(&kvm->srcu, idx);
+        if (is_error_page(aisb_page)) {
+            rc = -EIO;
+            goto unpin1;
+        }
+    }
+
Quoted 8 lines
Quoted 6 lines

+    /* AISB must be allocated before we can fill in GAITE */ +    mutex_lock(&aift->lock);
+    bit = airq_iv_alloc_bit(aift->sbv);
+    if (bit == -1UL)
+        goto unpin2;
+    zdev->aisb = bit;

aisb here is the aisb offset right?

Yes

Quoted 1 line

Then may be add a comment as in gait and fmt0 aisb is an address.

Sure, good point

Quoted 7 lines
Quoted 5 lines

+    zdev->aibv = airq_iv_create(msi_vecs, AIRQ_IV_DATA | +                          AIRQ_IV_BITLOCK |
+                          AIRQ_IV_GUESTVEC,
+                    (unsigned long *)fib->fmt0.aibv);

phys_to_virt ?

Ugh, yep -- we just put the physical address in fib->fmt0.aibv a few lines earlier via page_to_phys

Quoted 11 lines

arch/s390/kvm/pci.c int kvm_s390_pci_aen_init(u8 nisc) · line 259

Quoted 9 lines
arch/s390/kvm/pci.c +8 -0
+
+    spin_lock_irq(&aift->gait_lock);
+    gaite = (struct zpci_gaite *)aift->gait + (zdev->aisb *
+                           sizeof(struct zpci_gaite));
+
+    /* If assist not requested, host will get all alerts */
+    if (assist)
+        gaite->gisa = (u32)(u64)&kvm->arch.sie_page2->gisa;

virt_to_phys ?

Yes

Quoted 57 lines
Quoted 43 lines

+    else
+        gaite->gisa = 0;
+
+    gaite->gisc = fib->fmt0.isc;
+    gaite->count++;
+    gaite->aisbo = fib->fmt0.aisbo;
+    gaite->aisb = virt_to_phys(page_address(aisb_page) + (fib->fmt0.aisb &
+                                  ~PAGE_MASK));
+    aift->kzdev[zdev->aisb] = zdev->kzdev;
+    spin_unlock_irq(&aift->gait_lock);
+
+    /* Update guest FIB for re-issue */
+    fib->fmt0.aisbo = zdev->aisb & 63;
+    fib->fmt0.aisb = virt_to_phys(aift->sbv->vector + (zdev->aisb /
64) * 8);
+    fib->fmt0.isc = kvm_s390_gisc_register(kvm, gaite->gisc);

arch/s390/kvm/pci.c +25 -0
+
+    /* Save some guest fib values in the host for later use */
+    zdev->kzdev->fib.fmt0.isc = fib->fmt0.isc;
+    zdev->kzdev->fib.fmt0.aibv = fib->fmt0.aibv;
+    mutex_unlock(&aift->lock);
+
+    /* Issue the clp to setup the irq now */
+    rc = kvm_zpci_set_airq(zdev);
+    return rc;
+
+unpin2:
+    mutex_unlock(&aift->lock);
+    if (fib->fmt0.sum == 1) {
+        gaddr = page_to_phys(aisb_page);
+        kvm_release_pfn_dirty(gaddr >> PAGE_SHIFT);
+    }
+unpin1:
+    kvm_release_pfn_dirty(fib->fmt0.aibv >> PAGE_SHIFT);
+out:
+    return rc;
+}
+EXPORT_SYMBOL_GPL(kvm_s390_pci_aif_enable);
+
+int kvm_s390_pci_aif_disable(struct zpci_dev *zdev)
+{

arch/s390/kvm/pci.c int kvm_s390_pci_aen_init(u8 nisc) · line 309

Quoted 10 lines
arch/s390/kvm/pci.c +10 -0
+    struct kvm_zdev *kzdev = zdev->kzdev;
+    struct zpci_gaite *gaite;
+    int rc;
+    u8 isc;
+
+    if (zdev->gd == 0)
+        return -EINVAL;
+
+    /* Even if the clear fails due to an error, clear the GAITE */
+    rc = kvm_zpci_clear_airq(zdev);

Having a look at kvm_zpci_clear_airq() the only possible error seems to be when an error recovery is in progress.
The error returned for a wrong FH, function does not exist anymore, or if the interrupt vectors are already deregistered by the instruction are returned as success by the function.

Quoted 2 lines

How can we be sure that we have no conflict with a recovery in progress? Shouldn't we in this case let the recovery process handle the function and stop here?

Hmm -- So I think for a userspace-initiated call to this routine, yes. We could then assume recovery takes care of things. However, we also call this routine from vfio-pci core when closing the device...

So then let's look at how this would work -- the current recovery action for passthrough is always PCI_ERS_RESULT_DISCONNECT. The process of disconnecting the device will trigger vfio-pci to close it's device, which in turn will trigger vfio_pci_zdev_release() which will in turn also call kvm_390_aif_disable as part of cleanup. However, in this case now we want to clear the GAITE anyway even if kvm_zpci_clear_airq(zdev) fails now because we know the device is for sure going away.

I think I need some sort of input to this routine that indicates we must cleanup (bool force or something) which would only be specified by the call from vfio_pci_zdev_release().

Quoted 2 lines

Doesn't the aif lock mutex placed after and not before the clear_irq open a door for race condition with the recovery?

Good point.

Quoted 30 lines
Quoted 21 lines

+
+    mutex_lock(&aift->lock);
+    if (zdev->kzdev->fib.fmt0.aibv == 0)
+        goto out;
+    spin_lock_irq(&aift->gait_lock);
+    gaite = (struct zpci_gaite *)aift->gait + (zdev->aisb * +                           sizeof(struct zpci_gaite)); +    isc = gaite->gisc;
+    gaite->count--;
+    if (gaite->count == 0) {
+        /* Release guest AIBV and AISB */
+        kvm_release_pfn_dirty(kzdev->fib.fmt0.aibv >> PAGE_SHIFT); +        if (gaite->aisb != 0)
+            kvm_release_pfn_dirty(gaite->aisb >> PAGE_SHIFT); +        /* Clear the GAIT entry */
+        gaite->aisb = 0;
+        gaite->gisc = 0;
+        gaite->aisbo = 0;
+        gaite->gisa = 0;
+        aift->kzdev[zdev->aisb] = 0;

Quoted 8 lines

+        /* Clear zdev info */
+        airq_iv_free_bit(aift->sbv, zdev->aisb);
+        airq_iv_release(zdev->aibv);
+        zdev->aisb = 0;
+        zdev->aibv = NULL;
+    }
+    spin_unlock_irq(&aift->gait_lock);
+    kvm_s390_gisc_unregister(kzdev->kvm, isc);

Quoted 2 lines

Don't we need to check the return value?
And maybe to report it to the caller?

Well, actually, I think we really need to look at the kvm_s390_gisc_register() call during aif_enable -- I unconditionally assigned it to the fib when in fact it can also return a negative error value (which I never check for) -- so I will re-arrange the code in aif_enable() to do that earlier using a local variable and leave on error in aif_enable if this fails.

kvm_s390_gisc_register() returns 2 possible errors, which are shared with gisc_unregister -- So with that change we will detect these errors (not using GISA, bad guest ISC) at aif_enable time.

So then for gisc_unregister we should really only possibly hit the 3rd error (guest ISC is not registered). And if for some reason we hit that error at disable time, well, that's weird and unexpected (s390dbf?) but as far as userspace is concerned the GAITE is cleared and the gisc is unregistered, so I think we want to return success still to userspace. But we must do the checking at gisc_register() time and fail for the other cases there.

author maintainer
#56 lore raw

PATCH v2 20/30 KVM: s390: pci: provide routines for enabling/disabling IOAT assist

In reply to Matthew Rosato

These routines will be wired into the vfio_pci_zdev ioctl handlers to respond to requests to enable / disable a device for PCI I/O Address Translation assistance.

---
 arch/s390/include/asm/kvm_pci.h |  15 ++++
 arch/s390/include/asm/pci_dma.h |   2 +
 arch/s390/kvm/pci.c             | 139 ++++++++++++++++++++++++++++++++
 arch/s390/kvm/pci.h             |   2 +
 4 files changed, 158 insertions(+)
arch/s390/include/asm/kvm_pci.h +15 -0
index 01fe14fffd7a..770849f13a70 100644
@@ -16,11 +16,21 @@
 #include <linux/kvm_host.h>
 #include <linux/kvm.h>
 #include <linux/pci.h>
+#include <linux/mutex.h>
 #include <asm/pci_insn.h>
+#include <asm/pci_dma.h>
+
+struct kvm_zdev_ioat {
+	unsigned long *head[ZPCI_TABLE_PAGES];
+	unsigned long **seg;
+	unsigned long ***pt;
+	struct mutex lock;
+};
 
 struct kvm_zdev {
 	struct zpci_dev *zdev;
 	struct kvm *kvm;
+	struct kvm_zdev_ioat ioat;
 	struct zpci_fib fib;
 };
 
@@ -33,6 +43,11 @@ int kvm_s390_pci_aif_enable(struct zpci_dev *zdev, struct zpci_fib *fib,
 			    bool assist);
 int kvm_s390_pci_aif_disable(struct zpci_dev *zdev);
 
+int kvm_s390_pci_ioat_probe(struct zpci_dev *zdev);
+int kvm_s390_pci_ioat_enable(struct zpci_dev *zdev, u64 iota);
+int kvm_s390_pci_ioat_disable(struct zpci_dev *zdev);
+u8 kvm_s390_pci_get_dtsm(struct zpci_dev *zdev);
+
 int kvm_s390_pci_interp_probe(struct zpci_dev *zdev);
 int kvm_s390_pci_interp_enable(struct zpci_dev *zdev);
 int kvm_s390_pci_interp_disable(struct zpci_dev *zdev);
arch/s390/include/asm/pci_dma.h +2 -0
index 91e63426bdc5..69e616d0712c 100644
@@ -50,6 +50,8 @@ enum zpci_ioat_dtype {
 #define ZPCI_TABLE_ALIGN		ZPCI_TABLE_SIZE
 #define ZPCI_TABLE_ENTRY_SIZE		(sizeof(unsigned long))
 #define ZPCI_TABLE_ENTRIES		(ZPCI_TABLE_SIZE / ZPCI_TABLE_ENTRY_SIZE)
+#define ZPCI_TABLE_PAGES		(ZPCI_TABLE_SIZE >> PAGE_SHIFT)
+#define ZPCI_TABLE_ENTRIES_PAGES	(ZPCI_TABLE_ENTRIES * ZPCI_TABLE_PAGES)
 
 #define ZPCI_TABLE_BITS			11
 #define ZPCI_PT_BITS			8
arch/s390/kvm/pci.c +139 -0

175 lines not on this page · raw message

arch/s390/kvm/pci.h +2 -0
index 387b637863c9..54355634df82 100644
@@ -16,6 +16,8 @@
 #include <asm/airq.h>
 #include <asm/kvm_pci.h>
 
+#define KVM_S390_PCI_DTSM_MASK 0x40
+
 struct zpci_gaite {
 	u32 gisa;
 	u8 gisc;
Signature · 1 line
2.27.0
#57 lore raw

Re: [PATCH v2 20/30] KVM: s390: pci: provide routines for enabling/disabling IOAT assist

Quoted 19 lines · Matthew Rosato

These routines will be wired into the vfio_pci_zdev ioctl handlers to respond to requests to enable / disable a device for PCI I/O Address Translation assistance.

---
  arch/s390/include/asm/kvm_pci.h |  15 ++++
  arch/s390/include/asm/pci_dma.h |   2 +
  arch/s390/kvm/pci.c             | 139 ++++++++++++++++++++++++++++++++
  arch/s390/kvm/pci.h             |   2 +
  4 files changed, 158 insertions(+)
arch/s390/include/asm/kvm_pci.h +0 -0
index 01fe14fffd7a..770849f13a70 100644
@@ -16,11 +16,21 @@
  #include <linux/kvm_host.h>
  #include <linux/kvm.h>

arch/s390/include/asm/kvm_pci.h line 27

Quoted 10 lines
arch/s390/include/asm/kvm_pci.h +8 -0
  #include <linux/pci.h>
+#include <linux/mutex.h>
  #include <asm/pci_insn.h>
+#include <asm/pci_dma.h>
+
+struct kvm_zdev_ioat {
+	unsigned long *head[ZPCI_TABLE_PAGES];
+	unsigned long **seg;
+	unsigned long ***pt;
+	struct mutex lock;

Can we please rename the mutex ioat_lock to have a unique name easy to follow for maintenance. Can you please add a description about when the lock should be used?

Quoted 55 lines
arch/s390/include/asm/kvm_pci.h +7 -0
+};
 
  struct kvm_zdev {
  	struct zpci_dev *zdev;
  	struct kvm *kvm;
+	struct kvm_zdev_ioat ioat;
  	struct zpci_fib fib;
  };
 
@@ -33,6 +43,11 @@ int kvm_s390_pci_aif_enable(struct zpci_dev *zdev, struct zpci_fib *fib,
  			    bool assist);
  int kvm_s390_pci_aif_disable(struct zpci_dev *zdev);
 
+int kvm_s390_pci_ioat_probe(struct zpci_dev *zdev);
+int kvm_s390_pci_ioat_enable(struct zpci_dev *zdev, u64 iota);
+int kvm_s390_pci_ioat_disable(struct zpci_dev *zdev);
+u8 kvm_s390_pci_get_dtsm(struct zpci_dev *zdev);
+
  int kvm_s390_pci_interp_probe(struct zpci_dev *zdev);
  int kvm_s390_pci_interp_enable(struct zpci_dev *zdev);
  int kvm_s390_pci_interp_disable(struct zpci_dev *zdev);
arch/s390/include/asm/pci_dma.h +2 -0
index 91e63426bdc5..69e616d0712c 100644
@@ -50,6 +50,8 @@ enum zpci_ioat_dtype {
  #define ZPCI_TABLE_ALIGN		ZPCI_TABLE_SIZE
  #define ZPCI_TABLE_ENTRY_SIZE		(sizeof(unsigned long))
  #define ZPCI_TABLE_ENTRIES		(ZPCI_TABLE_SIZE / ZPCI_TABLE_ENTRY_SIZE)
+#define ZPCI_TABLE_PAGES		(ZPCI_TABLE_SIZE >> PAGE_SHIFT)
+#define ZPCI_TABLE_ENTRIES_PAGES	(ZPCI_TABLE_ENTRIES * ZPCI_TABLE_PAGES)
 
  #define ZPCI_TABLE_BITS			11
  #define ZPCI_PT_BITS			8
arch/s390/kvm/pci.c +3 -0
index 7ed9abc476b6..39c13c25a700 100644
@@ -13,12 +13,15 @@
  #include <asm/pci.h>
  #include <asm/pci_insn.h>
  #include <asm/pci_io.h>
+#include <asm/pci_dma.h>
  #include <asm/sclp.h>
  #include "pci.h"
  #include "kvm-s390.h"
 
  struct zpci_aift *aift;
 
+#define shadow_ioat_init zdev->kzdev->ioat.head[0]
+
  static inline int __set_irq_noiib(u16 ctl, u8 isc)
  {
  	union zpci_sic_iib iib = {{0}};

arch/s390/kvm/pci.c int kvm_s390_pci_aif_disable(struct zpci_dev *zdev) · line 352

Quoted 6 lines
arch/s390/kvm/pci.c +3 -0
@@ -344,6 +347,135 @@ int kvm_s390_pci_aif_disable(struct zpci_dev *zdev)
  }
  EXPORT_SYMBOL_GPL(kvm_s390_pci_aif_disable);
 
+int kvm_s390_pci_ioat_probe(struct zpci_dev *zdev)
+{
+	/* Must have a KVM association registered */

may be add something like : "The ioat structure is embeded in kzdev"

arch/s390/kvm/pci.c int kvm_s390_pci_aif_disable(struct zpci_dev *zdev) · line 353

Quoted 1 line
arch/s390/kvm/pci.c +1 -0
+	if (!zdev->kzdev || !zdev->kzdev->kvm)

Why do we need to check for kvm ?
Having kzdev is already tested by the unique caller.

Quoted 5 lines
arch/s390/kvm/pci.c +5 -0
+		return -EINVAL;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(kvm_s390_pci_ioat_probe);

arch/s390/kvm/pci.c int kvm_s390_pci_aif_disable(struct zpci_dev *zdev) · line 368

Quoted 10 lines
arch/s390/kvm/pci.c +10 -0
+
+int kvm_s390_pci_ioat_enable(struct zpci_dev *zdev, u64 iota)
+{
+	gpa_t gpa = (gpa_t)(iota & ZPCI_RTE_ADDR_MASK);
+	struct kvm_zdev_ioat *ioat;
+	struct page *page;
+	struct kvm *kvm;
+	unsigned int idx;
+	void *iaddr;
+	int i, rc = 0;

no need to initialize rc

Quoted 8 lines
arch/s390/kvm/pci.c +8 -0
+
+	if (shadow_ioat_init)
+		return -EINVAL;
+
+	/* Ensure supported type specified */
+	if ((iota & ZPCI_IOTA_RTTO_FLAG) != ZPCI_IOTA_RTTO_FLAG)
+		return -EINVAL;
+

arch/s390/kvm/pci.c int kvm_s390_pci_aif_disable(struct zpci_dev *zdev) · line 386

Quoted 10 lines
arch/s390/kvm/pci.c +10 -0
+	kvm = zdev->kzdev->kvm;
+	ioat = &zdev->kzdev->ioat;
+	mutex_lock(&ioat->lock);
+	idx = srcu_read_lock(&kvm->srcu);
+	for (i = 0; i < ZPCI_TABLE_PAGES; i++) {
+		page = gfn_to_page(kvm, gpa_to_gfn(gpa));
+		if (is_error_page(page)) {
+			srcu_read_unlock(&kvm->srcu, idx);
+			rc = -EIO;
+			goto out;

goto unpin ?

arch/s390/kvm/pci.c int kvm_s390_pci_aif_disable(struct zpci_dev *zdev) · line 395

Quoted 9 lines
arch/s390/kvm/pci.c +9 -0
+		}
+		iaddr = page_to_virt(page) + (gpa & ~PAGE_MASK);
+		ioat->head[i] = (unsigned long *)iaddr;
+		gpa += PAGE_SIZE;
+	}
+	srcu_read_unlock(&kvm->srcu, idx);
+
+	zdev->kzdev->ioat.seg = kcalloc(ZPCI_TABLE_ENTRIES_PAGES,
+					sizeof(unsigned long *), GFP_KERNEL);

What about:

        ioat->seg = kcalloc(ZPCI_TABLE_ENTRIES_PAGES,
                            sizeof(*ioat->seg), GFP_KERNEL);
	if (!ioat->seg)
...
	ioat->pt = ...

?

arch/s390/kvm/pci.c int kvm_s390_pci_aif_disable(struct zpci_dev *zdev) · line 405

Quoted 10 lines
arch/s390/kvm/pci.c +10 -0
+	if (!zdev->kzdev->ioat.seg)
+		goto unpin;
+	zdev->kzdev->ioat.pt = kcalloc(ZPCI_TABLE_ENTRIES,
+				       sizeof(unsigned long **), GFP_KERNEL);
+	if (!zdev->kzdev->ioat.pt)
+		goto free_seg;
+
+out:
+	mutex_unlock(&ioat->lock);
+	return rc;

return 0 ?

arch/s390/kvm/pci.c int kvm_s390_pci_aif_disable(struct zpci_dev *zdev) · line 408

Quoted 3 lines
arch/s390/kvm/pci.c +3 -0
+
+free_seg:
+	kfree(zdev->kzdev->ioat.seg);

kfree(ioat->seg) ?
rc = -ENOMEM;

arch/s390/kvm/pci.c int kvm_s390_pci_aif_disable(struct zpci_dev *zdev) · line 415

Quoted 7 lines
arch/s390/kvm/pci.c +7 -0
+unpin:
+	for (i = 0; i < ZPCI_TABLE_PAGES; i++) {
+		kvm_release_pfn_dirty((u64)ioat->head[i] >> PAGE_SHIFT);
+		ioat->head[i] = 0;
+	}
+	mutex_unlock(&ioat->lock);
+	return -ENOMEM;
	return rc;
Quoted 1 line
arch/s390/kvm/pci.c +1 -0
+}

...snip...

Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#58 lore raw

Re: [PATCH v2 20/30] KVM: s390: pci: provide routines for enabling/disabling IOAT assist

Quoted 27 lines · Pierre Morel
Quoted 24 lines · Matthew Rosato

These routines will be wired into the vfio_pci_zdev ioctl handlers to respond to requests to enable / disable a device for PCI I/O Address Translation assistance.

---
  arch/s390/include/asm/kvm_pci.h |  15 ++++ arch/s390/include/asm/pci_dma.h |   2 + arch/s390/kvm/pci.c             | 139 ++++++++++++++++++++++++++++++++ arch/s390/kvm/pci.h             |   2 +

  4 files changed, 158 insertions(+)
arch/s390/include/asm/kvm_pci.h +3 -0
index 01fe14fffd7a..770849f13a70 100644
@@ -16,11 +16,21 @@
   #include <linux/kvm_host.h>
   #include <linux/kvm.h>
   #include <linux/pci.h>
+#include <linux/mutex.h>
   #include <asm/pci_insn.h>
+#include <asm/pci_dma.h>
+
Quoted 9 lines

arch/s390/include/asm/kvm_pci.h line 27

Quoted 5 lines
arch/s390/include/asm/kvm_pci.h +5 -0
+struct kvm_zdev_ioat {
+    unsigned long *head[ZPCI_TABLE_PAGES];
+    unsigned long **seg;
+    unsigned long ***pt;
+    struct mutex lock;

Can we please rename the mutex ioat_lock to have a unique name easy to follow for maintenance. Can you please add a description about when the lock should be used?

OK. The lock is meant to protect the contents of kvm_zdev_ioat -- I'll think of something to describe it.

Quoted 59 lines
Quoted 51 lines
arch/s390/include/asm/kvm_pci.h +7 -0
+};
   struct kvm_zdev {
       struct zpci_dev *zdev;
       struct kvm *kvm;
+    struct kvm_zdev_ioat ioat;
       struct zpci_fib fib;
   };
@@ -33,6 +43,11 @@ int kvm_s390_pci_aif_enable(struct zpci_dev *zdev, 
 struct zpci_fib *fib,
                   bool assist);
   int kvm_s390_pci_aif_disable(struct zpci_dev *zdev);
+int kvm_s390_pci_ioat_probe(struct zpci_dev *zdev);
+int kvm_s390_pci_ioat_enable(struct zpci_dev *zdev, u64 iota);
+int kvm_s390_pci_ioat_disable(struct zpci_dev *zdev);
+u8 kvm_s390_pci_get_dtsm(struct zpci_dev *zdev);
+
   int kvm_s390_pci_interp_probe(struct zpci_dev *zdev);
   int kvm_s390_pci_interp_enable(struct zpci_dev *zdev);
   int kvm_s390_pci_interp_disable(struct zpci_dev *zdev);
arch/s390/include/asm/pci_dma.h +2 -0
index 91e63426bdc5..69e616d0712c 100644
@@ -50,6 +50,8 @@ enum zpci_ioat_dtype {
   #define ZPCI_TABLE_ALIGN        ZPCI_TABLE_SIZE
   #define ZPCI_TABLE_ENTRY_SIZE        (sizeof(unsigned long))
   #define ZPCI_TABLE_ENTRIES        (ZPCI_TABLE_SIZE / ZPCI_TABLE_ENTRY_SIZE)
+#define ZPCI_TABLE_PAGES        (ZPCI_TABLE_SIZE >> PAGE_SHIFT)
+#define ZPCI_TABLE_ENTRIES_PAGES    (ZPCI_TABLE_ENTRIES * 
 ZPCI_TABLE_PAGES)
   #define ZPCI_TABLE_BITS            11
   #define ZPCI_PT_BITS            8
arch/s390/kvm/pci.c +3 -0
index 7ed9abc476b6..39c13c25a700 100644
@@ -13,12 +13,15 @@
   #include <asm/pci.h>
   #include <asm/pci_insn.h>
   #include <asm/pci_io.h>
+#include <asm/pci_dma.h>
   #include <asm/sclp.h>
   #include "pci.h"
   #include "kvm-s390.h"
   struct zpci_aift *aift;
+#define shadow_ioat_init zdev->kzdev->ioat.head[0]
+
   static inline int __set_irq_noiib(u16 ctl, u8 isc)
   {
       union zpci_sic_iib iib = {{0}};

arch/s390/kvm/pci.c int kvm_s390_pci_aif_disable(struct zpci_dev *zdev) · line 351

Quoted 5 lines
arch/s390/kvm/pci.c +3 -0
@@ -344,6 +347,135 @@ int kvm_s390_pci_aif_disable(struct zpci_dev *zdev)
   }
   EXPORT_SYMBOL_GPL(kvm_s390_pci_aif_disable);
+int kvm_s390_pci_ioat_probe(struct zpci_dev *zdev)
+{
+    /* Must have a KVM association registered */

may be add something like : "The ioat structure is embeded in kzdev"

Quoted 5 lines

arch/s390/kvm/pci.c int kvm_s390_pci_aif_disable(struct zpci_dev *zdev) · line 353

Quoted 1 line
arch/s390/kvm/pci.c +1 -0
+    if (!zdev->kzdev || !zdev->kzdev->kvm)

Why do we need to check for kvm ?
Having kzdev is already tested by the unique caller.

We probably don't need to check for the kzdev because the caller already did this, agreed there.

But as for checking the kvm association, Alex asked for this in a comment to v1 (comment was against one of the vfio patches that call these routines) -- The reason being the probe comes from a userspace request and can be against any vfio-pci(-zdev) device at any time, and there's no point in proceeding if this device is not associated with a KVM guest -- It's possible for the KVM notifier to also pass a null KVM address -- so I think it's better to just be sure here. In a well-behaved environment we would never see this (so, another case for an s390dbf entry)

Quoted 9 lines
Quoted 9 lines
arch/s390/kvm/pci.c +9 -0
+        return -EINVAL;
+
+    return 0;
+}
+EXPORT_SYMBOL_GPL(kvm_s390_pci_ioat_probe);
+
+int kvm_s390_pci_ioat_enable(struct zpci_dev *zdev, u64 iota)
+{
+    gpa_t gpa = (gpa_t)(iota & ZPCI_RTE_ADDR_MASK);
Quoted 8 lines

arch/s390/kvm/pci.c int kvm_s390_pci_aif_disable(struct zpci_dev *zdev) · line 368

Quoted 6 lines
arch/s390/kvm/pci.c +6 -0
+    struct kvm_zdev_ioat *ioat;
+    struct page *page;
+    struct kvm *kvm;
+    unsigned int idx;
+    void *iaddr;
+    int i, rc = 0;

no need to initialize rc

Agree based on the changes below

Quoted 13 lines
Quoted 13 lines
arch/s390/kvm/pci.c +12 -0
+
+    if (shadow_ioat_init)
+        return -EINVAL;
+
+    /* Ensure supported type specified */
+    if ((iota & ZPCI_IOTA_RTTO_FLAG) != ZPCI_IOTA_RTTO_FLAG)
+        return -EINVAL;
+
+    kvm = zdev->kzdev->kvm;
+    ioat = &zdev->kzdev->ioat;
+    mutex_lock(&ioat->lock);
+    idx = srcu_read_lock(&kvm->srcu);
Quoted 8 lines

arch/s390/kvm/pci.c int kvm_s390_pci_aif_disable(struct zpci_dev *zdev) · line 386

Quoted 6 lines
arch/s390/kvm/pci.c +6 -0
+    for (i = 0; i < ZPCI_TABLE_PAGES; i++) {
+        page = gfn_to_page(kvm, gpa_to_gfn(gpa));
+        if (is_error_page(page)) {
+            srcu_read_unlock(&kvm->srcu, idx);
+            rc = -EIO;
+            goto out;

            goto unpin ?

Ah, right, in case we hit this error somewhere in the middle of the loop.

Quoted 11 lines

arch/s390/kvm/pci.c int kvm_s390_pci_aif_disable(struct zpci_dev *zdev) · line 395

Quoted 10 lines
arch/s390/kvm/pci.c +9 -0
+        }
+        iaddr = page_to_virt(page) + (gpa & ~PAGE_MASK);
+        ioat->head[i] = (unsigned long *)iaddr;
+        gpa += PAGE_SIZE;
+    }
+    srcu_read_unlock(&kvm->srcu, idx);
+
+    zdev->kzdev->ioat.seg = kcalloc(ZPCI_TABLE_ENTRIES_PAGES,
+                    sizeof(unsigned long *), GFP_KERNEL);
Quoted 8 lines

What about:

        ioat->seg = kcalloc(ZPCI_TABLE_ENTRIES_PAGES, sizeof(*ioat->seg), GFP_KERNEL); if (!ioat->seg)
...
    ioat->pt = ...
?

Yep, would be fine (seems I forgot about the local *ioat here)

Quoted 13 lines

arch/s390/kvm/pci.c int kvm_s390_pci_aif_disable(struct zpci_dev *zdev) · line 405

Quoted 11 lines
arch/s390/kvm/pci.c +10 -0
+    if (!zdev->kzdev->ioat.seg)
+        goto unpin;
+    zdev->kzdev->ioat.pt = kcalloc(ZPCI_TABLE_ENTRIES,
+                       sizeof(unsigned long **), GFP_KERNEL);
+    if (!zdev->kzdev->ioat.pt)
+        goto free_seg;
+
+out:
+    mutex_unlock(&ioat->lock);
+    return rc;

    return 0 ?

Yes, we can do that now that we don't goto out: after is_error_page

Quoted 16 lines

arch/s390/kvm/pci.c int kvm_s390_pci_aif_disable(struct zpci_dev *zdev) · line 408

Quoted 4 lines
arch/s390/kvm/pci.c +3 -0
+
+free_seg:
+    kfree(zdev->kzdev->ioat.seg);

kfree(ioat->seg) ?
rc = -ENOMEM;

arch/s390/kvm/pci.c int kvm_s390_pci_aif_disable(struct zpci_dev *zdev) · line 415

Quoted 7 lines
arch/s390/kvm/pci.c +7 -0
+unpin:
+    for (i = 0; i < ZPCI_TABLE_PAGES; i++) {
+        kvm_release_pfn_dirty((u64)ioat->head[i] >> PAGE_SHIFT);
+        ioat->head[i] = 0;
+    }
+    mutex_unlock(&ioat->lock);
+    return -ENOMEM;
Quoted 1 line

    return rc;

And yes, agreed, now that we come here for other reasons (-EIO) we must return rc here and also set rc=-ENOMEM as you say for the kfree case above.

#59 lore raw

Re: [PATCH v2 20/30] KVM: s390: pci: provide routines for enabling/disabling IOAT assist

Quoted 102 lines · Matthew Rosato
Quoted 27 lines · Pierre Morel
Quoted 24 lines · Matthew Rosato

These routines will be wired into the vfio_pci_zdev ioctl handlers to respond to requests to enable / disable a device for PCI I/O Address Translation assistance.

---
  arch/s390/include/asm/kvm_pci.h |  15 ++++ arch/s390/include/asm/pci_dma.h |   2 + arch/s390/kvm/pci.c             | 139 ++++++++++++++++++++++++++++++++ arch/s390/kvm/pci.h             |   2 +

  4 files changed, 158 insertions(+)
arch/s390/include/asm/kvm_pci.h +3 -0
index 01fe14fffd7a..770849f13a70 100644
@@ -16,11 +16,21 @@
   #include <linux/kvm_host.h>
   #include <linux/kvm.h>
   #include <linux/pci.h>
+#include <linux/mutex.h>
   #include <asm/pci_insn.h>
+#include <asm/pci_dma.h>
+
Quoted 9 lines

arch/s390/include/asm/kvm_pci.h line 27

Quoted 5 lines
arch/s390/include/asm/kvm_pci.h +5 -0
+struct kvm_zdev_ioat {
+    unsigned long *head[ZPCI_TABLE_PAGES];
+    unsigned long **seg;
+    unsigned long ***pt;
+    struct mutex lock;

Can we please rename the mutex ioat_lock to have a unique name easy to follow for maintenance. Can you please add a description about when the lock should be used?

OK.  The lock is meant to protect the contents of kvm_zdev_ioat -- I'll think of something to describe it.

Quoted 62 lines
Quoted 51 lines
arch/s390/include/asm/kvm_pci.h +7 -0
+};
   struct kvm_zdev {
       struct zpci_dev *zdev;
       struct kvm *kvm;
+    struct kvm_zdev_ioat ioat;
       struct zpci_fib fib;
   };
@@ -33,6 +43,11 @@ int kvm_s390_pci_aif_enable(struct zpci_dev *zdev, 
 struct zpci_fib *fib,
                   bool assist);
   int kvm_s390_pci_aif_disable(struct zpci_dev *zdev);
+int kvm_s390_pci_ioat_probe(struct zpci_dev *zdev);
+int kvm_s390_pci_ioat_enable(struct zpci_dev *zdev, u64 iota);
+int kvm_s390_pci_ioat_disable(struct zpci_dev *zdev);
+u8 kvm_s390_pci_get_dtsm(struct zpci_dev *zdev);
+
   int kvm_s390_pci_interp_probe(struct zpci_dev *zdev);
   int kvm_s390_pci_interp_enable(struct zpci_dev *zdev);
   int kvm_s390_pci_interp_disable(struct zpci_dev *zdev);
arch/s390/include/asm/pci_dma.h +2 -0
index 91e63426bdc5..69e616d0712c 100644
@@ -50,6 +50,8 @@ enum zpci_ioat_dtype {
   #define ZPCI_TABLE_ALIGN        ZPCI_TABLE_SIZE
   #define ZPCI_TABLE_ENTRY_SIZE        (sizeof(unsigned long))
   #define ZPCI_TABLE_ENTRIES        (ZPCI_TABLE_SIZE / ZPCI_TABLE_ENTRY_SIZE)
+#define ZPCI_TABLE_PAGES        (ZPCI_TABLE_SIZE >> PAGE_SHIFT)
+#define ZPCI_TABLE_ENTRIES_PAGES    (ZPCI_TABLE_ENTRIES * 
 ZPCI_TABLE_PAGES)
   #define ZPCI_TABLE_BITS            11
   #define ZPCI_PT_BITS            8
arch/s390/kvm/pci.c +3 -0
index 7ed9abc476b6..39c13c25a700 100644
@@ -13,12 +13,15 @@
   #include <asm/pci.h>
   #include <asm/pci_insn.h>
   #include <asm/pci_io.h>
+#include <asm/pci_dma.h>
   #include <asm/sclp.h>
   #include "pci.h"
   #include "kvm-s390.h"
   struct zpci_aift *aift;
+#define shadow_ioat_init zdev->kzdev->ioat.head[0]
+
   static inline int __set_irq_noiib(u16 ctl, u8 isc)
   {
       union zpci_sic_iib iib = {{0}};

arch/s390/kvm/pci.c int kvm_s390_pci_aif_disable(struct zpci_dev · line 352

Quoted 6 lines
arch/s390/kvm/pci.c +3 -0
@@ -344,6 +347,135 @@ int kvm_s390_pci_aif_disable(struct zpci_dev 
 *zdev)
   }
   EXPORT_SYMBOL_GPL(kvm_s390_pci_aif_disable);
+int kvm_s390_pci_ioat_probe(struct zpci_dev *zdev)
+{
+    /* Must have a KVM association registered */

may be add something like : "The ioat structure is embeded in kzdev"

Quoted 1 line
arch/s390/kvm/pci.c +1 -0
+    if (!zdev->kzdev || !zdev->kzdev->kvm)
Quoted 7 lines
Quoted 3 lines

Why do we need to check for kvm ?
Having kzdev is already tested by the unique caller.

We probably don't need to check for the kzdev because the caller already did this, agreed there.

But as for checking the kvm association, Alex asked for this in a comment to v1 (comment was against one of the vfio patches that call these routines) -- The reason being the probe comes from a userspace request and can be against any vfio-pci(-zdev) device at any time, and there's no point in proceeding if this device is not associated with a KVM guest -- It's possible for the KVM notifier to also pass a null KVM address -- so I think it's better to just be sure here.  In a well-behaved environment we would never see this (so, another case for an s390dbf entry)

I thought the check could be done even if the userspace is not associated with KVM. But of course OK if Alex asked I would have missed some point.

Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#60 lore raw

PATCH v2 21/30 KVM: s390: pci: handle refresh of PCI translations

In reply to Matthew Rosato

Add a routine that will perform a shadow operation between a guest and host IOAT. A subsequent patch will invoke this in response to an 04 RPCIT instruction intercept.

---
 arch/s390/include/asm/kvm_pci.h |   1 +
 arch/s390/include/asm/pci_dma.h |   1 +
 arch/s390/kvm/pci.c             | 208 +++++++++++++++++++++++++++++++-
 arch/s390/kvm/pci.h             |   8 +-
 4 files changed, 216 insertions(+), 2 deletions(-)
arch/s390/include/asm/kvm_pci.h +1 -0
index 770849f13a70..fa90729a35cf 100644
@@ -30,6 +30,7 @@ struct kvm_zdev_ioat {
 struct kvm_zdev {
 	struct zpci_dev *zdev;
 	struct kvm *kvm;
+	u64 rpcit_count;
 	struct kvm_zdev_ioat ioat;
 	struct zpci_fib fib;
 };
arch/s390/include/asm/pci_dma.h +1 -0
index 69e616d0712c..38004e0a4383 100644
@@ -52,6 +52,7 @@ enum zpci_ioat_dtype {
 #define ZPCI_TABLE_ENTRIES		(ZPCI_TABLE_SIZE / ZPCI_TABLE_ENTRY_SIZE)
 #define ZPCI_TABLE_PAGES		(ZPCI_TABLE_SIZE >> PAGE_SHIFT)
 #define ZPCI_TABLE_ENTRIES_PAGES	(ZPCI_TABLE_ENTRIES * ZPCI_TABLE_PAGES)
+#define ZPCI_TABLE_ENTRIES_PER_PAGE	(ZPCI_TABLE_ENTRIES / ZPCI_TABLE_PAGES)
 
 #define ZPCI_TABLE_BITS			11
 #define ZPCI_PT_BITS			8
arch/s390/kvm/pci.c +207 -1

224 lines not on this page · raw message

arch/s390/kvm/pci.h +7 -1
index 54355634df82..bb2be7fc3934 100644
@@ -18,6 +18,9 @@
 
 #define KVM_S390_PCI_DTSM_MASK 0x40
 
+#define KVM_S390_RPCIT_INS_RES 0x10
+#define KVM_S390_RPCIT_ERR 0x28
+
 struct zpci_gaite {
 	u32 gisa;
 	u8 gisc;
@@ -33,6 +36,7 @@ struct zpci_aift {
 	struct kvm_zdev **kzdev;
 	spinlock_t gait_lock; /* Protects the gait, used during AEN forward */
 	struct mutex lock; /* Protects the other structures in aift */
+	u32 mdd;
 };
 
 extern struct zpci_aift *aift;
@@ -47,7 +51,9 @@ static inline struct kvm *kvm_s390_pci_si_to_kvm(struct zpci_aift *aift,
 
 int kvm_s390_pci_aen_init(u8 nisc);
 void kvm_s390_pci_aen_exit(void);
-
+int kvm_s390_pci_refresh_trans(struct kvm_vcpu *vcpu, unsigned long req,
+			       unsigned long start, unsigned long end,
+			       u8 *status);
 int kvm_s390_pci_init(void);
 
 #endif /* __KVM_S390_PCI_H */
Signature · 1 line
2.27.0
#61 lore raw

Re: [PATCH v2 21/30] KVM: s390: pci: handle refresh of PCI translations

Quoted 188 lines · Matthew Rosato

Add a routine that will perform a shadow operation between a guest and host IOAT. A subsequent patch will invoke this in response to an 04 RPCIT instruction intercept.

---
  arch/s390/include/asm/kvm_pci.h |   1 +
  arch/s390/include/asm/pci_dma.h |   1 +
  arch/s390/kvm/pci.c             | 208 +++++++++++++++++++++++++++++++-
  arch/s390/kvm/pci.h             |   8 +-
  4 files changed, 216 insertions(+), 2 deletions(-)
arch/s390/include/asm/kvm_pci.h +1 -0
index 770849f13a70..fa90729a35cf 100644
@@ -30,6 +30,7 @@ struct kvm_zdev_ioat {
  struct kvm_zdev {
  	struct zpci_dev *zdev;
  	struct kvm *kvm;
+	u64 rpcit_count;
  	struct kvm_zdev_ioat ioat;
  	struct zpci_fib fib;
  };
arch/s390/include/asm/pci_dma.h +1 -0
index 69e616d0712c..38004e0a4383 100644
@@ -52,6 +52,7 @@ enum zpci_ioat_dtype {
  #define ZPCI_TABLE_ENTRIES		(ZPCI_TABLE_SIZE / ZPCI_TABLE_ENTRY_SIZE)
  #define ZPCI_TABLE_PAGES		(ZPCI_TABLE_SIZE >> PAGE_SHIFT)
  #define ZPCI_TABLE_ENTRIES_PAGES	(ZPCI_TABLE_ENTRIES * ZPCI_TABLE_PAGES)
+#define ZPCI_TABLE_ENTRIES_PER_PAGE	(ZPCI_TABLE_ENTRIES / ZPCI_TABLE_PAGES)
 
  #define ZPCI_TABLE_BITS			11
  #define ZPCI_PT_BITS			8
arch/s390/kvm/pci.c +144 -0
index 39c13c25a700..38d2b77ec565 100644
@@ -149,6 +149,208 @@ int kvm_s390_pci_aen_init(u8 nisc)
  	return rc;
  }
 
+static int dma_shadow_cpu_trans(struct kvm_vcpu *vcpu, unsigned long *entry,
+				unsigned long *gentry)
+{
+	phys_addr_t gaddr = 0;
+	unsigned long idx;
+	struct page *page;
+	kvm_pfn_t pfn;
+	gpa_t addr;
+	int rc = 0;
+
+	if (pt_entry_isvalid(*gentry)) {
+		/* pin and validate */
+		addr = *gentry & ZPCI_PTE_ADDR_MASK;
+		idx = srcu_read_lock(&vcpu->kvm->srcu);
+		page = gfn_to_page(vcpu->kvm, gpa_to_gfn(addr));
+		srcu_read_unlock(&vcpu->kvm->srcu, idx);
+		if (is_error_page(page))
+			return -EIO;
+		gaddr = page_to_phys(page) + (addr & ~PAGE_MASK);
+	}
+
+	if (pt_entry_isvalid(*entry)) {
+		/* Either we are invalidating, replacing or no-op */
+		if (gaddr != 0) {
+			if ((*entry & ZPCI_PTE_ADDR_MASK) == gaddr) {
+				/* Duplicate */
+				kvm_release_pfn_dirty(*entry >> PAGE_SHIFT);
+			} else {
+				/* Replace */
+				pfn = (*entry >> PAGE_SHIFT);
+				invalidate_pt_entry(entry);
+				set_pt_pfaa(entry, gaddr);
+				validate_pt_entry(entry);
+				kvm_release_pfn_dirty(pfn);
+				rc = 1;
+			}
+		} else {
+			/* Invalidate */
+			pfn = (*entry >> PAGE_SHIFT);
+			invalidate_pt_entry(entry);
+			kvm_release_pfn_dirty(pfn);
+			rc = 1;
+		}
+	} else if (gaddr != 0) {
+		/* New Entry */
+		set_pt_pfaa(entry, gaddr);
+		validate_pt_entry(entry);
+	}
+
+	return rc;
+}
+
+static unsigned long *dma_walk_guest_cpu_trans(struct kvm_vcpu *vcpu,
+					       struct kvm_zdev_ioat *ioat,
+					       dma_addr_t dma_addr)
+{
+	unsigned long *rto, *sto, *pto;
+	unsigned int rtx, rts, sx, px, idx;
+	struct page *page;
+	gpa_t addr;
+	int i;
+
+	/* Pin guest segment table if needed */
+	rtx = calc_rtx(dma_addr);
+	rto = ioat->head[(rtx / ZPCI_TABLE_ENTRIES_PER_PAGE)];
+	rts = rtx * ZPCI_TABLE_PAGES;
+	if (!ioat->seg[rts]) {
+		if (!reg_entry_isvalid(rto[rtx % ZPCI_TABLE_ENTRIES_PER_PAGE]))
+			return NULL;
+		sto = get_rt_sto(rto[rtx % ZPCI_TABLE_ENTRIES_PER_PAGE]);
+		addr = ((u64)sto & ZPCI_RTE_ADDR_MASK);
+		idx = srcu_read_lock(&vcpu->kvm->srcu);
+		for (i = 0; i < ZPCI_TABLE_PAGES; i++) {
+			page = gfn_to_page(vcpu->kvm, gpa_to_gfn(addr));
+			if (is_error_page(page)) {
+				srcu_read_unlock(&vcpu->kvm->srcu, idx);
+				return NULL;
+			}
+			ioat->seg[rts + i] = page_to_virt(page) +
+					     (addr & ~PAGE_MASK);
+			addr += PAGE_SIZE;
+		}
+		srcu_read_unlock(&vcpu->kvm->srcu, idx);
+	}
+
+	/* Allocate pin pointers for another segment table if needed */
+	if (!ioat->pt[rtx]) {
+		ioat->pt[rtx] = kcalloc(ZPCI_TABLE_ENTRIES,
+					(sizeof(unsigned long *)), GFP_KERNEL);
+		if (!ioat->pt[rtx])
+			return NULL;
+	}
+	/* Pin guest page table if needed */
+	sx = calc_sx(dma_addr);
+	sto = ioat->seg[(rts + (sx / ZPCI_TABLE_ENTRIES_PER_PAGE))];
+	if (!ioat->pt[rtx][sx]) {
+		if (!reg_entry_isvalid(sto[sx % ZPCI_TABLE_ENTRIES_PER_PAGE]))
+			return NULL;
+		pto = get_st_pto(sto[sx % ZPCI_TABLE_ENTRIES_PER_PAGE]);
+		if (!pto)
+			return NULL;
+		addr = ((u64)pto & ZPCI_STE_ADDR_MASK);
+		idx = srcu_read_lock(&vcpu->kvm->srcu);
+		page = gfn_to_page(vcpu->kvm, gpa_to_gfn(addr));
+		srcu_read_unlock(&vcpu->kvm->srcu, idx);
+		if (is_error_page(page))
+			return NULL;
+		ioat->pt[rtx][sx] = page_to_virt(page) + (addr & ~PAGE_MASK);
+	}
+	pto = ioat->pt[rtx][sx];
+
+	/* Return guest PTE */
+	px = calc_px(dma_addr);
+	return &pto[px];
+}
+
+
+static int dma_table_shadow(struct kvm_vcpu *vcpu, struct zpci_dev *zdev,
+			    dma_addr_t dma_addr, size_t size)
+{
+	unsigned int nr_pages = PAGE_ALIGN(size) >> PAGE_SHIFT;
+	struct kvm_zdev *kzdev = zdev->kzdev;
+	unsigned long *entry, *gentry;
+	int i, rc = 0, rc2;
+
+	if (!nr_pages || !kzdev)
+		return -EINVAL;
+
+	mutex_lock(&kzdev->ioat.lock);
+	if (!zdev->dma_table || !kzdev->ioat.head[0]) {
+		rc = -EINVAL;
+		goto out_unlock;
+	}
+
+	for (i = 0; i < nr_pages; i++) {
+		gentry = dma_walk_guest_cpu_trans(vcpu, &kzdev->ioat, dma_addr);
+		if (!gentry)
+			continue;
+		entry = dma_walk_cpu_trans(zdev->dma_table, dma_addr);
+
+		if (!entry) {
+			rc = -ENOMEM;
+			goto out_unlock;
+		}

arch/s390/kvm/pci.c int kvm_s390_pci_aen_init(u8 nisc) · line 305

Quoted 10 lines
arch/s390/kvm/pci.c +10 -0
+
+		rc2 = dma_shadow_cpu_trans(vcpu, entry, gentry);
+		if (rc2 < 0) {
+			rc = -EIO;
+			goto out_unlock;
+		}
+		dma_addr += PAGE_SIZE;
+		rc += rc2;
+	}
+

In case of error, shouldn't we invalidate the shadow tables entries we did validate until the error?

Quoted 7 lines
arch/s390/kvm/pci.c +7 -0
+out_unlock:
+	mutex_unlock(&kzdev->ioat.lock);
+	return rc;
+}
+
+int kvm_s390_pci_refresh_trans(struct kvm_vcpu *vcpu, unsigned long req,
+			       unsigned long start, unsigned long size,

arch/s390/kvm/pci.c int kvm_s390_pci_aen_init(u8 nisc) · line 322

Quoted 10 lines
arch/s390/kvm/pci.c +10 -0
+			       u8 *status)
+{
+	struct zpci_dev *zdev;
+	u32 fh = req >> 32;
+	int rc;
+
+	/* Make sure this is a valid device associated with this guest */
+	zdev = get_zdev_by_fh(fh);
+	if (!zdev || !zdev->kzdev || zdev->kzdev->kvm != vcpu->kvm) {
+		*status = 0;

Wouldn't it be interesting to add some debug information here. When would this appear?

Also if we have this error this looks like we have a VM problem, shouldn't we treat this in QEMU and return -EOPNOTSUPP ?

Quoted 86 lines
arch/s390/kvm/pci.c +36 -1
+		return -EINVAL;
+	}
+
+	/* Only proceed if the device is using the assist */
+	if (zdev->kzdev->ioat.head[0] == 0)
+		return -EOPNOTSUPP;
+
+	rc = dma_table_shadow(vcpu, zdev, start, size);
+	if (rc < 0) {
+		/*
+		 * If errors encountered during shadow operations, we must
+		 * fabricate status to present to the guest
+		 */
+		switch (rc) {
+		case -ENOMEM:
+			*status = KVM_S390_RPCIT_INS_RES;
+			break;
+		default:
+			*status = KVM_S390_RPCIT_ERR;
+			break;
+		}
+	} else if (rc > 0) {
+		/* Host RPCIT must be issued */
+		rc = zpci_refresh_trans((u64) zdev->fh << 32, start, size,
+					status);
+	}
+	zdev->kzdev->rpcit_count++;
+
+	return rc;
+}
+
  /* Modify PCI: Register floating adapter interruption forwarding */
  static int kvm_zpci_set_airq(struct zpci_dev *zdev)
  {
@@ -620,6 +822,8 @@ EXPORT_SYMBOL_GPL(kvm_s390_pci_attach_kvm);
 
  int kvm_s390_pci_init(void)
  {
+	int rc;
+
  	aift = kzalloc(sizeof(struct zpci_aift), GFP_KERNEL);
  	if (!aift)
  		return -ENOMEM;
@@ -627,5 +831,7 @@ int kvm_s390_pci_init(void)
  	spin_lock_init(&aift->gait_lock);
  	mutex_init(&aift->lock);
 
-	return 0;
+	rc = zpci_get_mdd(&aift->mdd);
+
+	return rc;
  }
arch/s390/kvm/pci.h +7 -1
index 54355634df82..bb2be7fc3934 100644
@@ -18,6 +18,9 @@
 
  #define KVM_S390_PCI_DTSM_MASK 0x40
 
+#define KVM_S390_RPCIT_INS_RES 0x10
+#define KVM_S390_RPCIT_ERR 0x28
+
  struct zpci_gaite {
  	u32 gisa;
  	u8 gisc;
@@ -33,6 +36,7 @@ struct zpci_aift {
  	struct kvm_zdev **kzdev;
  	spinlock_t gait_lock; /* Protects the gait, used during AEN forward */
  	struct mutex lock; /* Protects the other structures in aift */
+	u32 mdd;
  };
 
  extern struct zpci_aift *aift;
@@ -47,7 +51,9 @@ static inline struct kvm *kvm_s390_pci_si_to_kvm(struct zpci_aift *aift,
 
  int kvm_s390_pci_aen_init(u8 nisc);
  void kvm_s390_pci_aen_exit(void);
-
+int kvm_s390_pci_refresh_trans(struct kvm_vcpu *vcpu, unsigned long req,
+			       unsigned long start, unsigned long end,
+			       u8 *status);
  int kvm_s390_pci_init(void);
 
  #endif /* __KVM_S390_PCI_H */
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#62 lore raw

Re: [PATCH v2 21/30] KVM: s390: pci: handle refresh of PCI translations

Quoted 3 lines · Pierre Morel

On 1/14/22 21:31, Matthew Rosato wrote:

...

Quoted 32 lines
Quoted 32 lines

+static int dma_table_shadow(struct kvm_vcpu *vcpu, struct zpci_dev *zdev,
+                dma_addr_t dma_addr, size_t size)

arch/s390/kvm/pci.c +29 -0
+{
+    unsigned int nr_pages = PAGE_ALIGN(size) >> PAGE_SHIFT;
+    struct kvm_zdev *kzdev = zdev->kzdev;
+    unsigned long *entry, *gentry;
+    int i, rc = 0, rc2;
+
+    if (!nr_pages || !kzdev)
+        return -EINVAL;
+
+    mutex_lock(&kzdev->ioat.lock);
+    if (!zdev->dma_table || !kzdev->ioat.head[0]) {
+        rc = -EINVAL;
+        goto out_unlock;
+    }
+
+    for (i = 0; i < nr_pages; i++) {
+        gentry = dma_walk_guest_cpu_trans(vcpu, &kzdev->ioat, dma_addr);
+        if (!gentry)
+            continue;
+        entry = dma_walk_cpu_trans(zdev->dma_table, dma_addr);
+
+        if (!entry) {
+            rc = -ENOMEM;
+            goto out_unlock;
+        }
+
+        rc2 = dma_shadow_cpu_trans(vcpu, entry, gentry);
+        if (rc2 < 0) {
+            rc = -EIO;
Quoted 8 lines
Quoted 6 lines

+            goto out_unlock;
+        }
+        dma_addr += PAGE_SIZE;
+        rc += rc2;
+    }
+

In case of error, shouldn't we invalidate the shadow tables entries we did validate until the error?

Hmm, I don't think this is strictly necessary - the status returned should indicate the specified DMA range is now in an indeterminate state (putting the onus on the guest to take corrective action via a global refresh).

In fact I think I screwed that up below in kvm_s390_pci_refresh_trans, the fabricated status should always be KVM_S390_RPCIT_INS_RES.

Quoted 13 lines
Quoted 13 lines
arch/s390/kvm/pci.c +12 -0
+out_unlock:
+    mutex_unlock(&kzdev->ioat.lock);
+    return rc;
+}
+
+int kvm_s390_pci_refresh_trans(struct kvm_vcpu *vcpu, unsigned long req,
+                   unsigned long start, unsigned long size,
+                   u8 *status)
+{
+    struct zpci_dev *zdev;
+    u32 fh = req >> 32;
+    int rc;
Quoted 8 lines
Quoted 5 lines

+
+    /* Make sure this is a valid device associated with this guest */ +    zdev = get_zdev_by_fh(fh);
+    if (!zdev || !zdev->kzdev || zdev->kzdev->kvm != vcpu->kvm) { +        *status = 0;

Wouldn't it be interesting to add some debug information here. When would this appear?

Yes, I agree -- One of the follow-ons I'd like to add after this series is s390dbf entries; this seems like a good spot for one.

As to when this could happen; it should not under normal circumstances, but consider something like arbitrary function handles coming from the intercepted guest instruction. We need to ensure that the specified function 1) exists and 2) is associated with the guest issuing the refresh.

Quoted 3 lines

Also if we have this error this looks like we have a VM problem, shouldn't we treat this in QEMU and return -EOPNOTSUPP ?

Well, I'm not sure if we can really tell where the problem is (it could for example indicate a misbehaving guest, or a bug in our KVM tracking of hostdevs).

The guest chose the function handle, and if we got here then that means it doesn't indicate that it's an emulated device, which means either we are using the assist and KVM should handle the intercept or we are not and userspace should handle it. But in both of those cases, there should be a host device and it should be associated with the guest.

I think if we decide to throw this to userspace in this event, QEMU needs some extra code to handle it (basically, if QEMU receives the intercept and the device is neither emulated nor using intercept mode then we must treat as an invalid handle as this intercept should have been handled by KVM)

Quoted 12 lines
Quoted 12 lines

+        return -EINVAL;
+    }

arch/s390/kvm/pci.c +10 -0
+
+    /* Only proceed if the device is using the assist */
+    if (zdev->kzdev->ioat.head[0] == 0)
+        return -EOPNOTSUPP;
+
+    rc = dma_table_shadow(vcpu, zdev, start, size);
+    if (rc < 0) {
+        /*
+         * If errors encountered during shadow operations, we must
+         * fabricate status to present to the guest
Quoted 8 lines
Quoted 8 lines

+         */
+        switch (rc) {
+        case -ENOMEM:
+            *status = KVM_S390_RPCIT_INS_RES;
+            break;
+        default:
+            *status = KVM_S390_RPCIT_ERR;
+            break;

As mentioned above I think this switch statement should go away and instead always set KVM_S390_RPCIT_INS_RES when rc < 0.

Quoted 61 lines
Quoted 60 lines

+        }
+    } else if (rc > 0) {
+        /* Host RPCIT must be issued */
+        rc = zpci_refresh_trans((u64) zdev->fh << 32, start, size, +                    status);
+    }
+    zdev->kzdev->rpcit_count++;

arch/s390/kvm/pci.c +4 -0
+
+    return rc;
+}
+

  /* Modify PCI: Register floating adapter interruption forwarding */ static int kvm_zpci_set_airq(struct zpci_dev *zdev)
  {

arch/s390/kvm/pci.c +5 -1
@@ -620,6 +822,8 @@ EXPORT_SYMBOL_GPL(kvm_s390_pci_attach_kvm);
   int kvm_s390_pci_init(void)
   {
+    int rc;
+
       aift = kzalloc(sizeof(struct zpci_aift), GFP_KERNEL);
       if (!aift)
           return -ENOMEM;
@@ -627,5 +831,7 @@ int kvm_s390_pci_init(void)
       spin_lock_init(&aift->gait_lock);
       mutex_init(&aift->lock);
-    return 0;
+    rc = zpci_get_mdd(&aift->mdd);
+
+    return rc;
   }
arch/s390/kvm/pci.h +3 -0
index 54355634df82..bb2be7fc3934 100644
@@ -18,6 +18,9 @@
   #define KVM_S390_PCI_DTSM_MASK 0x40
+#define KVM_S390_RPCIT_INS_RES 0x10
+#define KVM_S390_RPCIT_ERR 0x28
+
   struct zpci_gaite {
       u32 gisa;
       u8 gisc;
@@ -33,6 +36,7 @@ struct zpci_aift {
       struct kvm_zdev **kzdev;

      spinlock_t gait_lock; /* Protects the gait, used during AEN forward */ struct mutex lock; /* Protects the other structures in aift */ +    u32 mdd;
  };
  extern struct zpci_aift *aift;

arch/s390/kvm/pci.h +3 -1
@@ -47,7 +51,9 @@ static inline struct kvm 
 *kvm_s390_pci_si_to_kvm(struct zpci_aift *aift,
   int kvm_s390_pci_aen_init(u8 nisc);
   void kvm_s390_pci_aen_exit(void);
-
+int kvm_s390_pci_refresh_trans(struct kvm_vcpu *vcpu, unsigned long req,
+                   unsigned long start, unsigned long end,
+                   u8 *status);
   int kvm_s390_pci_init(void);
   #endif /* __KVM_S390_PCI_H */
 
#63 lore raw

Re: [PATCH v2 21/30] KVM: s390: pci: handle refresh of PCI translations

Quoted 47 lines · Matthew Rosato
Quoted 3 lines · Pierre Morel

On 1/14/22 21:31, Matthew Rosato wrote:

...

Quoted 33 lines
Quoted 33 lines

+static int dma_table_shadow(struct kvm_vcpu *vcpu, struct zpci_dev *zdev,
+                dma_addr_t dma_addr, size_t size)

arch/s390/kvm/pci.c +17 -0
+{
+    unsigned int nr_pages = PAGE_ALIGN(size) >> PAGE_SHIFT;
+    struct kvm_zdev *kzdev = zdev->kzdev;
+    unsigned long *entry, *gentry;
+    int i, rc = 0, rc2;
+
+    if (!nr_pages || !kzdev)
+        return -EINVAL;
+
+    mutex_lock(&kzdev->ioat.lock);
+    if (!zdev->dma_table || !kzdev->ioat.head[0]) {
+        rc = -EINVAL;
+        goto out_unlock;
+    }
+
+    for (i = 0; i < nr_pages; i++) {
+        gentry = dma_walk_guest_cpu_trans(vcpu, &kzdev->ioat, 

dma_addr);
+        if (!gentry)
+            continue;
+        entry = dma_walk_cpu_trans(zdev->dma_table, dma_addr);

arch/s390/kvm/pci.c +9 -0
+
+        if (!entry) {
+            rc = -ENOMEM;
+            goto out_unlock;
+        }
+
+        rc2 = dma_shadow_cpu_trans(vcpu, entry, gentry);
+        if (rc2 < 0) {
+            rc = -EIO;
Quoted 8 lines
Quoted 6 lines

+            goto out_unlock;
+        }
+        dma_addr += PAGE_SIZE;
+        rc += rc2;
+    }
+

In case of error, shouldn't we invalidate the shadow tables entries we did validate until the error?

Quoted 3 lines

Hmm, I don't think this is strictly necessary - the status returned should indicate the specified DMA range is now in an indeterminate state (putting the onus on the guest to take corrective action via a global refresh).

In fact I think I screwed that up below in kvm_s390_pci_refresh_trans, the fabricated status should always be KVM_S390_RPCIT_INS_RES.

OK

Quoted 29 lines
Quoted 15 lines
Quoted 15 lines
arch/s390/kvm/pci.c +6 -0
+out_unlock:
+    mutex_unlock(&kzdev->ioat.lock);
+    return rc;
+}
+
+int kvm_s390_pci_refresh_trans(struct kvm_vcpu *vcpu, unsigned long 

req,
+                   unsigned long start, unsigned long size, +                   u8 *status)
+{
+    struct zpci_dev *zdev;
+    u32 fh = req >> 32;
+    int rc;

Quoted 8 lines
Quoted 5 lines

+
+    /* Make sure this is a valid device associated with this guest */ +    zdev = get_zdev_by_fh(fh);
+    if (!zdev || !zdev->kzdev || zdev->kzdev->kvm != vcpu->kvm) { +        *status = 0;

Wouldn't it be interesting to add some debug information here. When would this appear?

Yes, I agree -- One of the follow-ons I'd like to add after this series is s390dbf entries; this seems like a good spot for one.

As to when this could happen; it should not under normal circumstances, but consider something like arbitrary function handles coming from the intercepted guest instruction.  We need to ensure that the specified function 1) exists and 2) is associated with the guest issuing the refresh.

Quoted 1 line
Quoted 6 lines
Quoted 2 lines

Also if we have this error this looks like we have a VM problem, shouldn't we treat this in QEMU and return -EOPNOTSUPP ?

Well, I'm not sure if we can really tell where the problem is (it could for example indicate a misbehaving guest, or a bug in our KVM tracking of hostdevs).

The guest chose the function handle, and if we got here then that means it doesn't indicate that it's an emulated device, which means either we are using the assist and KVM should handle the intercept or we are not and userspace should handle it.  But in both of those cases, there should be a host device and it should be associated with the guest.

That is right if we can not find an associated zdev = F(fh) but the two other errors are KVM or QEMU errors AFAIU.

Quoted 2 lines

I think if we decide to throw this to userspace in this event, QEMU needs some extra code to handle it (basically, if QEMU receives the intercept and the device is neither emulated nor using intercept mode then we must treat as an invalid handle as this intercept should have been handled by KVM)

I do not want to start a discussion on this, I think we can let it like this at first and come back to it when we have a good idea on how to handle this. May be just add a /* TODO */

Quoted 88 lines
Quoted 14 lines
Quoted 14 lines

+        return -EINVAL;
+    }

arch/s390/kvm/pci.c +10 -0
+
+    /* Only proceed if the device is using the assist */
+    if (zdev->kzdev->ioat.head[0] == 0)
+        return -EOPNOTSUPP;
+
+    rc = dma_table_shadow(vcpu, zdev, start, size);
+    if (rc < 0) {
+        /*
+         * If errors encountered during shadow operations, we must
+         * fabricate status to present to the guest
Quoted 8 lines
Quoted 8 lines

+         */
+        switch (rc) {
+        case -ENOMEM:
+            *status = KVM_S390_RPCIT_INS_RES;
+            break;
+        default:
+            *status = KVM_S390_RPCIT_ERR;
+            break;

As mentioned above I think this switch statement should go away and instead always set KVM_S390_RPCIT_INS_RES when rc < 0.

Quoted 62 lines
Quoted 61 lines

+        }
+    } else if (rc > 0) {
+        /* Host RPCIT must be issued */
+        rc = zpci_refresh_trans((u64) zdev->fh << 32, start, size, +                    status);
+    }
+    zdev->kzdev->rpcit_count++;

arch/s390/kvm/pci.c +4 -0
+
+    return rc;
+}
+

  /* Modify PCI: Register floating adapter interruption forwarding */ static int kvm_zpci_set_airq(struct zpci_dev *zdev)
  {

arch/s390/kvm/pci.c +5 -1
@@ -620,6 +822,8 @@ EXPORT_SYMBOL_GPL(kvm_s390_pci_attach_kvm);
   int kvm_s390_pci_init(void)
   {
+    int rc;
+
       aift = kzalloc(sizeof(struct zpci_aift), GFP_KERNEL);
       if (!aift)
           return -ENOMEM;
@@ -627,5 +831,7 @@ int kvm_s390_pci_init(void)
       spin_lock_init(&aift->gait_lock);
       mutex_init(&aift->lock);
-    return 0;
+    rc = zpci_get_mdd(&aift->mdd);
+
+    return rc;
   }
arch/s390/kvm/pci.h +3 -0
index 54355634df82..bb2be7fc3934 100644
@@ -18,6 +18,9 @@
   #define KVM_S390_PCI_DTSM_MASK 0x40
+#define KVM_S390_RPCIT_INS_RES 0x10
+#define KVM_S390_RPCIT_ERR 0x28
+
   struct zpci_gaite {
       u32 gisa;
       u8 gisc;
@@ -33,6 +36,7 @@ struct zpci_aift {
       struct kvm_zdev **kzdev;

      spinlock_t gait_lock; /* Protects the gait, used during AEN forward */ struct mutex lock; /* Protects the other structures in aift */ +    u32 mdd;
  };
  extern struct zpci_aift *aift;

arch/s390/kvm/pci.h +3 -1
@@ -47,7 +51,9 @@ static inline struct kvm 
 *kvm_s390_pci_si_to_kvm(struct zpci_aift *aift,
   int kvm_s390_pci_aen_init(u8 nisc);
   void kvm_s390_pci_aen_exit(void);
-
+int kvm_s390_pci_refresh_trans(struct kvm_vcpu *vcpu, unsigned long 
 req,
+                   unsigned long start, unsigned long end,
+                   u8 *status);
   int kvm_s390_pci_init(void);
   #endif /* __KVM_S390_PCI_H */
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#64 lore raw

Re: [PATCH v2 21/30] KVM: s390: pci: handle refresh of PCI translations

Quoted 88 lines · Pierre Morel
Quoted 47 lines · Matthew Rosato
Quoted 3 lines · Pierre Morel

On 1/14/22 21:31, Matthew Rosato wrote:

...

Quoted 33 lines
Quoted 33 lines

+static int dma_table_shadow(struct kvm_vcpu *vcpu, struct zpci_dev *zdev,
+                dma_addr_t dma_addr, size_t size)

arch/s390/kvm/pci.c +17 -0
+{
+    unsigned int nr_pages = PAGE_ALIGN(size) >> PAGE_SHIFT;
+    struct kvm_zdev *kzdev = zdev->kzdev;
+    unsigned long *entry, *gentry;
+    int i, rc = 0, rc2;
+
+    if (!nr_pages || !kzdev)
+        return -EINVAL;
+
+    mutex_lock(&kzdev->ioat.lock);
+    if (!zdev->dma_table || !kzdev->ioat.head[0]) {
+        rc = -EINVAL;
+        goto out_unlock;
+    }
+
+    for (i = 0; i < nr_pages; i++) {
+        gentry = dma_walk_guest_cpu_trans(vcpu, &kzdev->ioat, 

dma_addr);
+        if (!gentry)
+            continue;
+        entry = dma_walk_cpu_trans(zdev->dma_table, dma_addr);

arch/s390/kvm/pci.c +9 -0
+
+        if (!entry) {
+            rc = -ENOMEM;
+            goto out_unlock;
+        }
+
+        rc2 = dma_shadow_cpu_trans(vcpu, entry, gentry);
+        if (rc2 < 0) {
+            rc = -EIO;
Quoted 8 lines
Quoted 6 lines

+            goto out_unlock;
+        }
+        dma_addr += PAGE_SIZE;
+        rc += rc2;
+    }
+

In case of error, shouldn't we invalidate the shadow tables entries we did validate until the error?

Quoted 3 lines

Hmm, I don't think this is strictly necessary - the status returned should indicate the specified DMA range is now in an indeterminate state (putting the onus on the guest to take corrective action via a global refresh).

In fact I think I screwed that up below in kvm_s390_pci_refresh_trans, the fabricated status should always be KVM_S390_RPCIT_INS_RES.

OK

Quoted 32 lines
Quoted 15 lines
Quoted 15 lines
arch/s390/kvm/pci.c +6 -0
+out_unlock:
+    mutex_unlock(&kzdev->ioat.lock);
+    return rc;
+}
+
+int kvm_s390_pci_refresh_trans(struct kvm_vcpu *vcpu, unsigned long 

req,
+                   unsigned long start, unsigned long size, +                   u8 *status)
+{
+    struct zpci_dev *zdev;
+    u32 fh = req >> 32;
+    int rc;

Quoted 8 lines
Quoted 5 lines

+
+    /* Make sure this is a valid device associated with this guest */ +    zdev = get_zdev_by_fh(fh);
+    if (!zdev || !zdev->kzdev || zdev->kzdev->kvm != vcpu->kvm) { +        *status = 0;

Wouldn't it be interesting to add some debug information here. When would this appear?

Yes, I agree -- One of the follow-ons I'd like to add after this series is s390dbf entries; this seems like a good spot for one.

As to when this could happen; it should not under normal circumstances, but consider something like arbitrary function handles coming from the intercepted guest instruction.  We need to ensure that the specified function 1) exists and 2) is associated with the guest issuing the refresh.

Quoted 3 lines

Also if we have this error this looks like we have a VM problem, shouldn't we treat this in QEMU and return -EOPNOTSUPP ?

Quoted 6 lines
Quoted 3 lines

Well, I'm not sure if we can really tell where the problem is (it could for example indicate a misbehaving guest, or a bug in our KVM tracking of hostdevs).

The guest chose the function handle, and if we got here then that means it doesn't indicate that it's an emulated device, which means either we are using the assist and KVM should handle the intercept or we are not and userspace should handle it.  But in both of those cases, there should be a host device and it should be associated with the guest.

That is right if we can not find an associated zdev = F(fh) but the two other errors are KVM or QEMU errors AFAIU.

I don't think we know for sure for any of the cases... For a well-behaved guest I agree with your assessment. However, the guest decides what fh to put into its refresh instruction and so a misbehaving guest could just pick arbitrary numbers for fh and circumstantially match some other host device. What if the guest just decided to try every single possible fh number in a loop with a refresh instruction? That's neither KVM nor QEMU's fault but can trip each of these cases.

Consider the different cases:

!zdev - Either the guest provided a bogus fh, KVM provided a bad fh via the VFIO ioctl which then QEMU fed into CLP or KVM provided the right fh via ioctl but QEMU clobbered it when providing it to the guest via CLP.

!zdev->kzdev - Either the guest provided a bogus fh that just so happened to match a host fh that has no KVM association, or KVM or QEMU screwed up somewhere (as above or because we failed to make the KVM assocation somehow)

kzdev->kvm != vcpu->kvm - Pretty much the same as above, but the matching device is actually in use by some other guest. Again it's possible the a misbehaving guest 'got lucky' with an arbitrary fh that happened to match a host fh with an existing KVM association -- or more likely that KVM or QEMU screwed up somewhere.

Quoted 6 lines
Quoted 3 lines

I think if we decide to throw this to userspace in this event, QEMU needs some extra code to handle it (basically, if QEMU receives the intercept and the device is neither emulated nor using intercept mode then we must treat as an invalid handle as this intercept should have been handled by KVM)

I do not want to start a discussion on this, I think we can let it like this at first and come back to it when we have a good idea on how to handle this. May be just add a /* TODO */

OK, sure. In any of the above cases, we are certainly done in KVM anyway. Whether there's value in passing it onto userspace vs immediately giving an error, let's think about it.

#65 lore raw

Re: [PATCH v2 21/30] KVM: s390: pci: handle refresh of PCI translations

Quoted 98 lines · Matthew Rosato
Quoted 88 lines · Pierre Morel
Quoted 47 lines · Matthew Rosato
Quoted 3 lines · Pierre Morel

On 1/14/22 21:31, Matthew Rosato wrote:

...

Quoted 33 lines
Quoted 33 lines

+static int dma_table_shadow(struct kvm_vcpu *vcpu, struct zpci_dev *zdev,
+                dma_addr_t dma_addr, size_t size)

arch/s390/kvm/pci.c +17 -0
+{
+    unsigned int nr_pages = PAGE_ALIGN(size) >> PAGE_SHIFT;
+    struct kvm_zdev *kzdev = zdev->kzdev;
+    unsigned long *entry, *gentry;
+    int i, rc = 0, rc2;
+
+    if (!nr_pages || !kzdev)
+        return -EINVAL;
+
+    mutex_lock(&kzdev->ioat.lock);
+    if (!zdev->dma_table || !kzdev->ioat.head[0]) {
+        rc = -EINVAL;
+        goto out_unlock;
+    }
+
+    for (i = 0; i < nr_pages; i++) {
+        gentry = dma_walk_guest_cpu_trans(vcpu, &kzdev->ioat, 

dma_addr);
+        if (!gentry)
+            continue;
+        entry = dma_walk_cpu_trans(zdev->dma_table, dma_addr);

arch/s390/kvm/pci.c +9 -0
+
+        if (!entry) {
+            rc = -ENOMEM;
+            goto out_unlock;
+        }
+
+        rc2 = dma_shadow_cpu_trans(vcpu, entry, gentry);
+        if (rc2 < 0) {
+            rc = -EIO;
Quoted 8 lines
Quoted 6 lines

+            goto out_unlock;
+        }
+        dma_addr += PAGE_SIZE;
+        rc += rc2;
+    }
+

In case of error, shouldn't we invalidate the shadow tables entries we did validate until the error?

Quoted 3 lines

Hmm, I don't think this is strictly necessary - the status returned should indicate the specified DMA range is now in an indeterminate state (putting the onus on the guest to take corrective action via a global refresh).

In fact I think I screwed that up below in kvm_s390_pci_refresh_trans, the fabricated status should always be KVM_S390_RPCIT_INS_RES.

OK

Quoted 32 lines
Quoted 15 lines
Quoted 15 lines
arch/s390/kvm/pci.c +6 -0
+out_unlock:
+    mutex_unlock(&kzdev->ioat.lock);
+    return rc;
+}
+
+int kvm_s390_pci_refresh_trans(struct kvm_vcpu *vcpu, unsigned 

long req,
+                   unsigned long start, unsigned long size, +                   u8 *status)
+{
+    struct zpci_dev *zdev;
+    u32 fh = req >> 32;
+    int rc;

Quoted 8 lines
Quoted 5 lines

+
+    /* Make sure this is a valid device associated with this guest */ +    zdev = get_zdev_by_fh(fh);
+    if (!zdev || !zdev->kzdev || zdev->kzdev->kvm != vcpu->kvm) { +        *status = 0;

Wouldn't it be interesting to add some debug information here. When would this appear?

Yes, I agree -- One of the follow-ons I'd like to add after this series is s390dbf entries; this seems like a good spot for one.

As to when this could happen; it should not under normal circumstances, but consider something like arbitrary function handles coming from the intercepted guest instruction.  We need to ensure that the specified function 1) exists and 2) is associated with the guest issuing the refresh.

Quoted 3 lines

Also if we have this error this looks like we have a VM problem, shouldn't we treat this in QEMU and return -EOPNOTSUPP ?

Quoted 6 lines
Quoted 3 lines

Well, I'm not sure if we can really tell where the problem is (it could for example indicate a misbehaving guest, or a bug in our KVM tracking of hostdevs).

The guest chose the function handle, and if we got here then that means it doesn't indicate that it's an emulated device, which means either we are using the assist and KVM should handle the intercept or we are not and userspace should handle it.  But in both of those cases, there should be a host device and it should be associated with the guest.

That is right if we can not find an associated zdev = F(fh) but the two other errors are KVM or QEMU errors AFAIU.

I don't think we know for sure for any of the cases...  For a well-behaved guest I agree with your assessment.  However, the guest decides what fh to put into its refresh instruction and so a misbehaving guest could just pick arbitrary numbers for fh and circumstantially match some other host device.  What if the guest just decided to try every single possible fh number in a loop with a refresh instruction? That's neither KVM nor QEMU's fault but can trip each of these cases.

Quoted 7 lines

Consider the different cases:

!zdev - Either the guest provided a bogus fh, KVM provided a bad fh via the VFIO ioctl which then QEMU fed into CLP or KVM provided the right fh via ioctl but QEMU clobbered it when providing it to the guest via CLP.

!zdev->kzdev - Either the guest provided a bogus fh that just so happened to match a host fh that has no KVM association, or KVM or QEMU screwed up somewhere (as above or because we failed to make the KVM assocation somehow)

kzdev->kvm != vcpu->kvm - Pretty much the same as above, but the matching device is actually in use by some other guest.  Again it's possible the a misbehaving guest 'got lucky' with an arbitrary fh that happened to match a host fh with an existing KVM association -- or more likely that KVM or QEMU screwed up somewhere.

OK, I understand and you are right, my error was to consider that get_zdev_by_fh() returns a zdev associated with a valid FH for the guest while it returns a zdev associated with a valid FH for the host.

If the comment would have been after the get_zdev_by_fh() and before the test I may be wouldn't have done this mistake.

Quoted 9 lines
Quoted 7 lines
Quoted 4 lines

I think if we decide to throw this to userspace in this event, QEMU needs some extra code to handle it (basically, if QEMU receives the intercept and the device is neither emulated nor using intercept mode then we must treat as an invalid handle as this intercept should have been handled by KVM)

I do not want to start a discussion on this, I think we can let it like this at first and come back to it when we have a good idea on how to handle this. May be just add a /* TODO */

OK, sure.  In any of the above cases, we are certainly done in KVM anyway.  Whether there's value in passing it onto userspace vs immediately giving an error, let's think about it.

No, I do not think we should anymore.
Sorry for this wrong idea.

Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#66 lore raw

PATCH v2 22/30 KVM: s390: intercept the rpcit instruction

In reply to Matthew Rosato

For faster handling of PCI translation refreshes, intercept in KVM and call the associated handler.

---
 arch/s390/kvm/priv.c | 46 ++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 46 insertions(+)
arch/s390/kvm/priv.c +46 -0
index 417154b314a6..5b65c1830de2 100644
@@ -29,6 +29,7 @@
 #include <asm/ap.h>
 #include "gaccess.h"
 #include "kvm-s390.h"
+#include "pci.h"
 #include "trace.h"
 
 static int handle_ri(struct kvm_vcpu *vcpu)
@@ -335,6 +336,49 @@ static int handle_rrbe(struct kvm_vcpu *vcpu)
 	return 0;
 }
 
+static int handle_rpcit(struct kvm_vcpu *vcpu)
+{
+	int reg1, reg2;
+	u8 status;
+	int rc;
+
+	if (vcpu->arch.sie_block->gpsw.mask & PSW_MASK_PSTATE)
+		return kvm_s390_inject_program_int(vcpu, PGM_PRIVILEGED_OP);
+
+	/* If the host doesn't support PCI, it must be an emulated device */
+	if (!IS_ENABLED(CONFIG_PCI))
+		return -EOPNOTSUPP;
+
+	kvm_s390_get_regs_rre(vcpu, &reg1, &reg2);
+
+	/* If the device has a SHM bit on, let userspace take care of this */
+	if (((vcpu->run->s.regs.gprs[reg1] >> 32) & aift->mdd) != 0)
+		return -EOPNOTSUPP;
+
+	rc = kvm_s390_pci_refresh_trans(vcpu, vcpu->run->s.regs.gprs[reg1],
+					vcpu->run->s.regs.gprs[reg2],
+					vcpu->run->s.regs.gprs[reg2+1],
+					&status);
+
+	switch (rc) {
+	case 0:
+		kvm_s390_set_psw_cc(vcpu, 0);
+		break;
+	case -EOPNOTSUPP:
+		return -EOPNOTSUPP;
+	default:
+		vcpu->run->s.regs.gprs[reg1] &= 0xffffffff00ffffffUL;
+		vcpu->run->s.regs.gprs[reg1] |= (u64) status << 24;
+		if (status != 0)
+			kvm_s390_set_psw_cc(vcpu, 1);
+		else
+			kvm_s390_set_psw_cc(vcpu, 3);
+		break;
+	}
+
+	return 0;
+}
+
 #define SSKE_NQ 0x8
 #define SSKE_MR 0x4
 #define SSKE_MC 0x2
@@ -1275,6 +1319,8 @@ int kvm_s390_handle_b9(struct kvm_vcpu *vcpu)
 		return handle_essa(vcpu);
 	case 0xaf:
 		return handle_pfmf(vcpu);
+	case 0xd3:
+		return handle_rpcit(vcpu);
 	default:
 		return -EOPNOTSUPP;
 	}
Signature · 1 line
2.27.0
#67 lore raw

Re: [PATCH v2 22/30] KVM: s390: intercept the rpcit instruction

Quoted 26 lines · Matthew Rosato

For faster handling of PCI translation refreshes, intercept in KVM and call the associated handler.

---
  arch/s390/kvm/priv.c | 46 ++++++++++++++++++++++++++++++++++++++++++++
  1 file changed, 46 insertions(+)
arch/s390/kvm/priv.c +3 -0
index 417154b314a6..5b65c1830de2 100644
@@ -29,6 +29,7 @@
  #include <asm/ap.h>
  #include "gaccess.h"
  #include "kvm-s390.h"
+#include "pci.h"
  #include "trace.h"
 
  static int handle_ri(struct kvm_vcpu *vcpu)
@@ -335,6 +336,49 @@ static int handle_rrbe(struct kvm_vcpu *vcpu)
  	return 0;
  }
 
+static int handle_rpcit(struct kvm_vcpu *vcpu)
+{

arch/s390/kvm/priv.c static int handle_rrbe(struct kvm_vcpu *vcpu) · line 350

Quoted 10 lines
arch/s390/kvm/priv.c +10 -0
+	int reg1, reg2;
+	u8 status;
+	int rc;
+
+	if (vcpu->arch.sie_block->gpsw.mask & PSW_MASK_PSTATE)
+		return kvm_s390_inject_program_int(vcpu, PGM_PRIVILEGED_OP);
+
+	/* If the host doesn't support PCI, it must be an emulated device */
+	if (!IS_ENABLED(CONFIG_PCI))
+		return -EOPNOTSUPP;

AFAIU this makes also sure that the following code is not compiled in case PCI is not supported.

I am not very used to compilation options, is it true with all our compilers and options? Or do we have to specify a compiler version?

Another concern is, shouldn't we use IS_ENABLED(CONFIG_VFIO_PCI) ?

Quoted 44 lines
arch/s390/kvm/priv.c +33 -0
+
+	kvm_s390_get_regs_rre(vcpu, &reg1, &reg2);
+
+	/* If the device has a SHM bit on, let userspace take care of this */
+	if (((vcpu->run->s.regs.gprs[reg1] >> 32) & aift->mdd) != 0)
+		return -EOPNOTSUPP;
+
+	rc = kvm_s390_pci_refresh_trans(vcpu, vcpu->run->s.regs.gprs[reg1],
+					vcpu->run->s.regs.gprs[reg2],
+					vcpu->run->s.regs.gprs[reg2+1],
+					&status);
+
+	switch (rc) {
+	case 0:
+		kvm_s390_set_psw_cc(vcpu, 0);
+		break;
+	case -EOPNOTSUPP:
+		return -EOPNOTSUPP;
+	default:
+		vcpu->run->s.regs.gprs[reg1] &= 0xffffffff00ffffffUL;
+		vcpu->run->s.regs.gprs[reg1] |= (u64) status << 24;
+		if (status != 0)
+			kvm_s390_set_psw_cc(vcpu, 1);
+		else
+			kvm_s390_set_psw_cc(vcpu, 3);
+		break;
+	}
+
+	return 0;
+}
+
  #define SSKE_NQ 0x8
  #define SSKE_MR 0x4
  #define SSKE_MC 0x2
@@ -1275,6 +1319,8 @@ int kvm_s390_handle_b9(struct kvm_vcpu *vcpu)
  		return handle_essa(vcpu);
  	case 0xaf:
  		return handle_pfmf(vcpu);
+	case 0xd3:
+		return handle_rpcit(vcpu);
  	default:
  		return -EOPNOTSUPP;
  	}
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#68 lore raw

Re: [PATCH v2 22/30] KVM: s390: intercept the rpcit instruction

Quoted 40 lines · Pierre Morel
Quoted 24 lines · Matthew Rosato

For faster handling of PCI translation refreshes, intercept in KVM and call the associated handler.

---
  arch/s390/kvm/priv.c | 46 ++++++++++++++++++++++++++++++++++++++++++++

  1 file changed, 46 insertions(+)
arch/s390/kvm/priv.c +3 -0
index 417154b314a6..5b65c1830de2 100644
@@ -29,6 +29,7 @@
   #include <asm/ap.h>
   #include "gaccess.h"
   #include "kvm-s390.h"
+#include "pci.h"
   #include "trace.h"
   static int handle_ri(struct kvm_vcpu *vcpu)
@@ -335,6 +336,49 @@ static int handle_rrbe(struct kvm_vcpu *vcpu)
       return 0;
   }
+static int handle_rpcit(struct kvm_vcpu *vcpu)
+{

arch/s390/kvm/priv.c static int handle_rrbe(struct kvm_vcpu *vcpu) · line 349

Quoted 10 lines
arch/s390/kvm/priv.c +10 -0
+    int reg1, reg2;
+    u8 status;
+    int rc;
+
+    if (vcpu->arch.sie_block->gpsw.mask & PSW_MASK_PSTATE)
+        return kvm_s390_inject_program_int(vcpu, PGM_PRIVILEGED_OP);
+
+    /* If the host doesn't support PCI, it must be an emulated device */
+    if (!IS_ENABLED(CONFIG_PCI))
+        return -EOPNOTSUPP;

AFAIU this makes also sure that the following code is not compiled in case PCI is not supported.

Quoted 4 lines

I am not very used to compilation options, is it true with all our compilers and options? Or do we have to specify a compiler version?

Another concern is, shouldn't we use IS_ENABLED(CONFIG_VFIO_PCI) ?

Same idea as in the other thread -- What we are trying to protect against here is referencing symbols that won't be linked (like zpci_refresh_trans, or the aift->mdd a few lines below)

It is indeed true that we should never need to handle the rpcit intercept in KVM if CONFIG_VFIO_PCI=n -- but the necessary symbols/code are linked at least, so we can just let the SHM logic sort this out. When CONFIG_PCI=y|m, arch/s390/kvm/pci.o will be linked and so we can compare the function handle against afit->mdd (check to see if the device is emulated) and use this to determine whether or not to immediately send to userspace -- And if CONFIG_VFIO_PCI=n, a SHM bit will always be on and so we'll always go to userspace via this check.

Quoted 49 lines
Quoted 48 lines
arch/s390/kvm/priv.c +4 -0
+
+    kvm_s390_get_regs_rre(vcpu, &reg1, &reg2);
+
+    /* If the device has a SHM bit on, let userspace take care of 

this */
+    if (((vcpu->run->s.regs.gprs[reg1] >> 32) & aift->mdd) != 0) +        return -EOPNOTSUPP;

arch/s390/kvm/priv.c +25 -0
+
+    rc = kvm_s390_pci_refresh_trans(vcpu, vcpu->run->s.regs.gprs[reg1],
+                    vcpu->run->s.regs.gprs[reg2],
+                    vcpu->run->s.regs.gprs[reg2+1],
+                    &status);
+
+    switch (rc) {
+    case 0:
+        kvm_s390_set_psw_cc(vcpu, 0);
+        break;
+    case -EOPNOTSUPP:
+        return -EOPNOTSUPP;
+    default:
+        vcpu->run->s.regs.gprs[reg1] &= 0xffffffff00ffffffUL;
+        vcpu->run->s.regs.gprs[reg1] |= (u64) status << 24;
+        if (status != 0)
+            kvm_s390_set_psw_cc(vcpu, 1);
+        else
+            kvm_s390_set_psw_cc(vcpu, 3);
+        break;
+    }
+
+    return 0;
+}
+

  #define SSKE_NQ 0x8
  #define SSKE_MR 0x4
  #define SSKE_MC 0x2

arch/s390/kvm/priv.c +0 -0
@@ -1275,6 +1319,8 @@ int kvm_s390_handle_b9(struct kvm_vcpu *vcpu)
           return handle_essa(vcpu);

      case 0xaf:
          return handle_pfmf(vcpu);
+    case 0xd3:
+        return handle_rpcit(vcpu);
      default:
          return -EOPNOTSUPP;
      }

#69 lore raw

Re: [PATCH v2 22/30] KVM: s390: intercept the rpcit instruction

Quoted 42 lines · Matthew Rosato
Quoted 41 lines · Pierre Morel
Quoted 32 lines · Matthew Rosato

For faster handling of PCI translation refreshes, intercept in KVM and call the associated handler.

---
  arch/s390/kvm/priv.c | 46 ++++++++++++++++++++++++++++++++++++++++++++

  1 file changed, 46 insertions(+)
arch/s390/kvm/priv.c +11 -0
index 417154b314a6..5b65c1830de2 100644
@@ -29,6 +29,7 @@
   #include <asm/ap.h>
   #include "gaccess.h"
   #include "kvm-s390.h"
+#include "pci.h"
   #include "trace.h"
   static int handle_ri(struct kvm_vcpu *vcpu)
@@ -335,6 +336,49 @@ static int handle_rrbe(struct kvm_vcpu *vcpu)
       return 0;
   }
+static int handle_rpcit(struct kvm_vcpu *vcpu)
+{
+    int reg1, reg2;
+    u8 status;
+    int rc;
+
+    if (vcpu->arch.sie_block->gpsw.mask & PSW_MASK_PSTATE)
+        return kvm_s390_inject_program_int(vcpu, PGM_PRIVILEGED_OP);
+
+    /* If the host doesn't support PCI, it must be an emulated 
Quoted 3 lines

device */
+    if (!IS_ENABLED(CONFIG_PCI))
+        return -EOPNOTSUPP;

AFAIU this makes also sure that the following code is not compiled in case PCI is not supported.

Quoted 8 lines
Quoted 4 lines

I am not very used to compilation options, is it true with all our compilers and options? Or do we have to specify a compiler version?

Another concern is, shouldn't we use IS_ENABLED(CONFIG_VFIO_PCI) ?

Same idea as in the other thread -- What we are trying to protect against here is referencing symbols that won't be linked (like zpci_refresh_trans, or the aift->mdd a few lines below)

It is indeed true that we should never need to handle the rpcit intercept in KVM if CONFIG_VFIO_PCI=n -- but the necessary symbols/code are linked at least, so we can just let the SHM logic sort this out. When CONFIG_PCI=y|m, arch/s390/kvm/pci.o will be linked and so we can compare the function handle against afit->mdd (check to see if the device is emulated) and use this to determine whether or not to immediately send to userspace -- And if CONFIG_VFIO_PCI=n, a SHM bit will always be on and so we'll always go to userspace via this check.

So we agree.
But as I I said somewhere else I wonder if CONFIG_VFIO_PCI_ZDEV would not even be better here.

Quoted 51 lines
Quoted 50 lines
Quoted 49 lines
arch/s390/kvm/priv.c +4 -0
+
+    kvm_s390_get_regs_rre(vcpu, &reg1, &reg2);
+
+    /* If the device has a SHM bit on, let userspace take care of 

this */
+    if (((vcpu->run->s.regs.gprs[reg1] >> 32) & aift->mdd) != 0) +        return -EOPNOTSUPP;

arch/s390/kvm/priv.c +25 -0
+
+    rc = kvm_s390_pci_refresh_trans(vcpu, vcpu->run->s.regs.gprs[reg1],
+                    vcpu->run->s.regs.gprs[reg2],
+                    vcpu->run->s.regs.gprs[reg2+1],
+                    &status);
+
+    switch (rc) {
+    case 0:
+        kvm_s390_set_psw_cc(vcpu, 0);
+        break;
+    case -EOPNOTSUPP:
+        return -EOPNOTSUPP;
+    default:
+        vcpu->run->s.regs.gprs[reg1] &= 0xffffffff00ffffffUL;
+        vcpu->run->s.regs.gprs[reg1] |= (u64) status << 24;
+        if (status != 0)
+            kvm_s390_set_psw_cc(vcpu, 1);
+        else
+            kvm_s390_set_psw_cc(vcpu, 3);
+        break;
+    }
+
+    return 0;
+}
+

  #define SSKE_NQ 0x8
  #define SSKE_MR 0x4
  #define SSKE_MC 0x2

arch/s390/kvm/priv.c +0 -0
@@ -1275,6 +1319,8 @@ int kvm_s390_handle_b9(struct kvm_vcpu *vcpu)
           return handle_essa(vcpu);

      case 0xaf:
          return handle_pfmf(vcpu);
+    case 0xd3:
+        return handle_rpcit(vcpu);
      default:
          return -EOPNOTSUPP;
      }

Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
#70 lore raw

Re: [PATCH v2 22/30] KVM: s390: intercept the rpcit instruction

In reply to Matthew Rosato
Quoted 4 lines · Matthew Rosato

For faster handling of PCI translation refreshes, intercept in KVM and call the associated handler.

Aside our previous discussion, 2 small codingstyle to fix

Quoted 32 lines
---
  arch/s390/kvm/priv.c | 46 ++++++++++++++++++++++++++++++++++++++++++++
  1 file changed, 46 insertions(+)
arch/s390/kvm/priv.c +13 -0
index 417154b314a6..5b65c1830de2 100644
@@ -29,6 +29,7 @@
  #include <asm/ap.h>
  #include "gaccess.h"
  #include "kvm-s390.h"
+#include "pci.h"
  #include "trace.h"
 
  static int handle_ri(struct kvm_vcpu *vcpu)
@@ -335,6 +336,49 @@ static int handle_rrbe(struct kvm_vcpu *vcpu)
  	return 0;
  }
 
+static int handle_rpcit(struct kvm_vcpu *vcpu)
+{
+	int reg1, reg2;
+	u8 status;
+	int rc;
+
+	if (vcpu->arch.sie_block->gpsw.mask & PSW_MASK_PSTATE)
+		return kvm_s390_inject_program_int(vcpu, PGM_PRIVILEGED_OP);
+
+	/* If the host doesn't support PCI, it must be an emulated device */
+	if (!IS_ENABLED(CONFIG_PCI))
+		return -EOPNOTSUPP;

arch/s390/kvm/priv.c static int handle_rrbe(struct kvm_vcpu *vcpu) · line 360

Quoted 10 lines
arch/s390/kvm/priv.c +10 -0
+
+	kvm_s390_get_regs_rre(vcpu, &reg1, &reg2);
+
+	/* If the device has a SHM bit on, let userspace take care of this */
+	if (((vcpu->run->s.regs.gprs[reg1] >> 32) & aift->mdd) != 0)
+		return -EOPNOTSUPP;
+
+	rc = kvm_s390_pci_refresh_trans(vcpu, vcpu->run->s.regs.gprs[reg1],
+					vcpu->run->s.regs.gprs[reg2],
+					vcpu->run->s.regs.gprs[reg2+1],

Here, spaces around "+"

arch/s390/kvm/priv.c static int handle_rrbe(struct kvm_vcpu *vcpu) · line 371

Quoted 11 lines
arch/s390/kvm/priv.c +11 -0
+					&status);
+
+	switch (rc) {
+	case 0:
+		kvm_s390_set_psw_cc(vcpu, 0);
+		break;
+	case -EOPNOTSUPP:
+		return -EOPNOTSUPP;
+	default:
+		vcpu->run->s.regs.gprs[reg1] &= 0xffffffff00ffffffUL;
+		vcpu->run->s.regs.gprs[reg1] |= (u64) status << 24;

Here no blank after cast.

Quoted 23 lines
arch/s390/kvm/priv.c +12 -0
+		if (status != 0)
+			kvm_s390_set_psw_cc(vcpu, 1);
+		else
+			kvm_s390_set_psw_cc(vcpu, 3);
+		break;
+	}
+
+	return 0;
+}
+
  #define SSKE_NQ 0x8
  #define SSKE_MR 0x4
  #define SSKE_MC 0x2
@@ -1275,6 +1319,8 @@ int kvm_s390_handle_b9(struct kvm_vcpu *vcpu)
  		return handle_essa(vcpu);
  	case 0xaf:
  		return handle_pfmf(vcpu);
+	case 0xd3:
+		return handle_rpcit(vcpu);
  	default:
  		return -EOPNOTSUPP;
  	}
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#71 lore raw

PATCH v2 23/30 vfio/pci: re-introduce CONFIG_VFIO_PCI_ZDEV

In reply to Matthew Rosato

This was previously removed as unnecessary; while that was true, subsequent changes will make KVM an additional required component for vfio-pci-zdev. Let's re-introduce CONFIG_VFIO_PCI_ZDEV as now there is actually a reason to say 'n' for it (when not planning to CONFIG_KVM).

---
 drivers/vfio/pci/Kconfig      | 11 +++++++++++
 drivers/vfio/pci/Makefile     |  2 +-
 include/linux/vfio_pci_core.h |  2 +-
 3 files changed, 13 insertions(+), 2 deletions(-)
drivers/vfio/pci/Kconfig +11 -0
index 860424ccda1b..fedd1d4cb592 100644
@@ -42,5 +42,16 @@ config VFIO_PCI_IGD
 	  and LPC bridge config space.
 
 	  To enable Intel IGD assignment through vfio-pci, say Y.
+
+config VFIO_PCI_ZDEV
+	bool "VFIO PCI extensions for s390x KVM passthrough"
+	depends on S390 && KVM
+	default y
+	help
+	  Support s390x-specific extensions to enable support for enhancements
+	  to KVM passthrough capabilities, such as interpretive execution of
+	  zPCI instructions.
+
+	  To enable s390x KVM vfio-pci extensions, say Y.
 endif
 endif
drivers/vfio/pci/Makefile +1 -1
index 349d68d242b4..01b1f83d83d7 100644
@@ -1,7 +1,7 @@
 # SPDX-License-Identifier: GPL-2.0-only
 
 vfio-pci-core-y := vfio_pci_core.o vfio_pci_intrs.o vfio_pci_rdwr.o vfio_pci_config.o
-vfio-pci-core-$(CONFIG_S390) += vfio_pci_zdev.o
+vfio-pci-core-$(CONFIG_VFIO_PCI_ZDEV) += vfio_pci_zdev.o
 obj-$(CONFIG_VFIO_PCI_CORE) += vfio-pci-core.o
 
 vfio-pci-y := vfio_pci.o
include/linux/vfio_pci_core.h +1 -1
index ef9a44b6cf5d..5e2bca3b89db 100644
@@ -195,7 +195,7 @@ static inline int vfio_pci_igd_init(struct vfio_pci_core_device *vdev)
 }
 #endif
 
-#ifdef CONFIG_S390
+#ifdef CONFIG_VFIO_PCI_ZDEV
 extern int vfio_pci_info_zdev_add_caps(struct vfio_pci_core_device *vdev,
 				       struct vfio_info_cap *caps);
 #else
Signature · 1 line
2.27.0
#72 lore raw

Re: [PATCH v2 23/30] vfio/pci: re-introduce CONFIG_VFIO_PCI_ZDEV

Quoted 21 lines · Matthew Rosato

This was previously removed as unnecessary; while that was true, subsequent changes will make KVM an additional required component for vfio-pci-zdev. Let's re-introduce CONFIG_VFIO_PCI_ZDEV as now there is actually a reason to say 'n' for it (when not planning to CONFIG_KVM).

---
  drivers/vfio/pci/Kconfig      | 11 +++++++++++
  drivers/vfio/pci/Makefile     |  2 +-
  include/linux/vfio_pci_core.h |  2 +-
  3 files changed, 13 insertions(+), 2 deletions(-)
drivers/vfio/pci/Kconfig +1 -0
index 860424ccda1b..fedd1d4cb592 100644
@@ -42,5 +42,16 @@ config VFIO_PCI_IGD
  	  and LPC bridge config space.
 
  	  To enable Intel IGD assignment through vfio-pci, say Y.
+

drivers/vfio/pci/Kconfig config VFIO_PCI_IGD · line 55

Quoted 10 lines
drivers/vfio/pci/Kconfig +10 -0
+config VFIO_PCI_ZDEV
+	bool "VFIO PCI extensions for s390x KVM passthrough"
+	depends on S390 && KVM
+	default y
+	help
+	  Support s390x-specific extensions to enable support for enhancements
+	  to KVM passthrough capabilities, such as interpretive execution of
+	  zPCI instructions.
+
+	  To enable s390x KVM vfio-pci extensions, say Y.

In several patches we check on CONFIG_PCI (14,15,16,17 and 22) but we may have PCI without VFIO_PCI, wouldn't it be a problem?

Here we define a new CONFIG entry and I have two questions:

1- there is no dependency on VFIO_PCI while the functionality is obviously based on VFIO_PCI

2- Wouldn't it be possible to use this item and the single condition for the different checks we need through the new VFIO interpretation functionality.

Quoted 29 lines
drivers/vfio/pci/Kconfig +0 -0
  endif
  endif
drivers/vfio/pci/Makefile +1 -1
index 349d68d242b4..01b1f83d83d7 100644
@@ -1,7 +1,7 @@
  # SPDX-License-Identifier: GPL-2.0-only
 
  vfio-pci-core-y := vfio_pci_core.o vfio_pci_intrs.o vfio_pci_rdwr.o vfio_pci_config.o
-vfio-pci-core-$(CONFIG_S390) += vfio_pci_zdev.o
+vfio-pci-core-$(CONFIG_VFIO_PCI_ZDEV) += vfio_pci_zdev.o
  obj-$(CONFIG_VFIO_PCI_CORE) += vfio-pci-core.o
 
  vfio-pci-y := vfio_pci.o
include/linux/vfio_pci_core.h +1 -1
index ef9a44b6cf5d..5e2bca3b89db 100644
@@ -195,7 +195,7 @@ static inline int vfio_pci_igd_init(struct vfio_pci_core_device *vdev)
  }
  #endif
 
-#ifdef CONFIG_S390
+#ifdef CONFIG_VFIO_PCI_ZDEV
  extern int vfio_pci_info_zdev_add_caps(struct vfio_pci_core_device *vdev,
  				       struct vfio_info_cap *caps);
  #else
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#73 lore raw

Re: [PATCH v2 23/30] vfio/pci: re-introduce CONFIG_VFIO_PCI_ZDEV

Quoted 37 lines · Pierre Morel
Quoted 26 lines · Matthew Rosato

This was previously removed as unnecessary; while that was true, subsequent changes will make KVM an additional required component for vfio-pci-zdev. Let's re-introduce CONFIG_VFIO_PCI_ZDEV as now there is actually a reason to say 'n' for it (when not planning to CONFIG_KVM).

---
  drivers/vfio/pci/Kconfig      | 11 +++++++++++
  drivers/vfio/pci/Makefile     |  2 +-
  include/linux/vfio_pci_core.h |  2 +-

  3 files changed, 13 insertions(+), 2 deletions(-)
drivers/vfio/pci/Kconfig +0 -0
index 860424ccda1b..fedd1d4cb592 100644
@@ -42,5 +42,16 @@ config VFIO_PCI_IGD

        and LPC bridge config space.
        To enable Intel IGD assignment through vfio-pci, say Y.

drivers/vfio/pci/Kconfig +7 -0
+
+config VFIO_PCI_ZDEV
+    bool "VFIO PCI extensions for s390x KVM passthrough"
+    depends on S390 && KVM
+    default y
+    help
+      Support s390x-specific extensions to enable support for 
Quoted 5 lines

enhancements
+      to KVM passthrough capabilities, such as interpretive execution of +      zPCI instructions.
+
+      To enable s390x KVM vfio-pci extensions, say Y.

In several patches we check on CONFIG_PCI (14,15,16,17 and 22) but we may have PCI without VFIO_PCI, wouldn't it be a problem?

Quoted 3 lines

Here we define a new CONFIG entry and I have two questions:

1- there is no dependency on VFIO_PCI while the functionality is obviously based on VFIO_PCI

It's not obvious from this diff, but this 'config VFIO_PCI_ZDEV' statement is within an 'if VFIO_PCI' statement, just like VFIO_PCI_IGD above -- so the dependency is there.

Quoted 2 lines

2- Wouldn't it be possible to use this item and the single condition for the different checks we need through the new VFIO interpretation functionality.

Possibly, but 1) we'd have to make linking arch/s390/kvm/pci.o dependent on CONFIG_VFIO_PCI instead of CONFIG_PCI in patch 14 and 2) if the relationship between CONFIG_VFIO_PCI and CONFIG_PCI were to ever change (though I don't see why it would..), we would be broken because the symbols we are referencing really require CONFIG_PCI (as they are located in s390 PCI).

#74 lore raw

Re: [PATCH v2 23/30] vfio/pci: re-introduce CONFIG_VFIO_PCI_ZDEV

Quoted 39 lines · Matthew Rosato
Quoted 38 lines · Pierre Morel
Quoted 26 lines · Matthew Rosato

This was previously removed as unnecessary; while that was true, subsequent changes will make KVM an additional required component for vfio-pci-zdev. Let's re-introduce CONFIG_VFIO_PCI_ZDEV as now there is actually a reason to say 'n' for it (when not planning to CONFIG_KVM).

---
  drivers/vfio/pci/Kconfig      | 11 +++++++++++
  drivers/vfio/pci/Makefile     |  2 +-
  include/linux/vfio_pci_core.h |  2 +-

  3 files changed, 13 insertions(+), 2 deletions(-)
drivers/vfio/pci/Kconfig +0 -0
index 860424ccda1b..fedd1d4cb592 100644
@@ -42,5 +42,16 @@ config VFIO_PCI_IGD

        and LPC bridge config space.
        To enable Intel IGD assignment through vfio-pci, say Y.

drivers/vfio/pci/Kconfig +7 -0
+
+config VFIO_PCI_ZDEV
+    bool "VFIO PCI extensions for s390x KVM passthrough"
+    depends on S390 && KVM
+    default y
+    help
+      Support s390x-specific extensions to enable support for 
Quoted 6 lines

enhancements
+      to KVM passthrough capabilities, such as interpretive execution of
+      zPCI instructions.
+
+      To enable s390x KVM vfio-pci extensions, say Y.

In several patches we check on CONFIG_PCI (14,15,16,17 and 22) but we may have PCI without VFIO_PCI, wouldn't it be a problem?

Quoted 5 lines
Quoted 3 lines

Here we define a new CONFIG entry and I have two questions:

1- there is no dependency on VFIO_PCI while the functionality is obviously based on VFIO_PCI

It's not obvious from this diff, but this 'config VFIO_PCI_ZDEV' statement is within an 'if VFIO_PCI' statement, just like VFIO_PCI_IGD above -- so the dependency is there.

sorry, I remember now you already answered this to Christian last time.

Quoted 6 lines
Quoted 3 lines

2- Wouldn't it be possible to use this item and the single condition for the different checks we need through the new VFIO interpretation functionality.

Possibly, but 1) we'd have to make linking arch/s390/kvm/pci.o dependent on CONFIG_VFIO_PCI instead of CONFIG_PCI in patch 14 and 2) if the relationship between CONFIG_VFIO_PCI and CONFIG_PCI were to ever change (though I don't see why it would..), we would be broken because the symbols we are referencing really require CONFIG_PCI (as they are located in s390 PCI).

Yes but VFIO_PCI_ZDEV depends on KVM, PCI and on VFIO_PCI
Wouldn't a single config item for this new code be easier to manage and understand?

Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#75 lore raw

Re: [PATCH v2 23/30] vfio/pci: re-introduce CONFIG_VFIO_PCI_ZDEV

Quoted 52 lines · Pierre Morel
Quoted 39 lines · Matthew Rosato
Quoted 38 lines · Pierre Morel
Quoted 26 lines · Matthew Rosato

This was previously removed as unnecessary; while that was true, subsequent changes will make KVM an additional required component for vfio-pci-zdev. Let's re-introduce CONFIG_VFIO_PCI_ZDEV as now there is actually a reason to say 'n' for it (when not planning to CONFIG_KVM).

---
  drivers/vfio/pci/Kconfig      | 11 +++++++++++
  drivers/vfio/pci/Makefile     |  2 +-
  include/linux/vfio_pci_core.h |  2 +-

  3 files changed, 13 insertions(+), 2 deletions(-)
drivers/vfio/pci/Kconfig +0 -0
index 860424ccda1b..fedd1d4cb592 100644
@@ -42,5 +42,16 @@ config VFIO_PCI_IGD

        and LPC bridge config space.
        To enable Intel IGD assignment through vfio-pci, say Y.

drivers/vfio/pci/Kconfig +7 -0
+
+config VFIO_PCI_ZDEV
+    bool "VFIO PCI extensions for s390x KVM passthrough"
+    depends on S390 && KVM
+    default y
+    help
+      Support s390x-specific extensions to enable support for 
Quoted 6 lines

enhancements
+      to KVM passthrough capabilities, such as interpretive execution of
+      zPCI instructions.
+
+      To enable s390x KVM vfio-pci extensions, say Y.

In several patches we check on CONFIG_PCI (14,15,16,17 and 22) but we may have PCI without VFIO_PCI, wouldn't it be a problem?

Quoted 5 lines
Quoted 3 lines

Here we define a new CONFIG entry and I have two questions:

1- there is no dependency on VFIO_PCI while the functionality is obviously based on VFIO_PCI

It's not obvious from this diff, but this 'config VFIO_PCI_ZDEV' statement is within an 'if VFIO_PCI' statement, just like VFIO_PCI_IGD above -- so the dependency is there.

sorry, I remember now you already answered this to Christian last time.

Quoted 2 lines
Quoted 2 lines
Quoted 8 lines
Quoted 4 lines
Quoted 1 line

2- Wouldn't it be possible to use this item and the single condition for the different checks we need through the new VFIO interpretation functionality.

Possibly, but 1) we'd have to make linking arch/s390/kvm/pci.o dependent on CONFIG_VFIO_PCI instead of CONFIG_PCI in patch 14 and 2) if the relationship between CONFIG_VFIO_PCI and CONFIG_PCI were to ever change (though I don't see why it would..), we would be broken because the symbols we are referencing really require CONFIG_PCI (as they are located in s390 PCI).

Yes but VFIO_PCI_ZDEV depends on KVM, PCI and on VFIO_PCI
Wouldn't a single config item for this new code be easier to manage and understand?

I guess my primary resistance is to abstracting/hiding the dependency. Yes, userspace will never setup for zPCI interpretation without CONFIG_VFIO_PCI{_ZDEV}, but that's not where the compilation dependency is -- it's on CONFIG_PCI specifically.

But I guess on the other hand you could argue why even bother building pci.o into kvm without CONFIG_VFIO_PCI_ZDEV as it will never be used.

OK, I will have a look at making this change. It will require a little reorganization, at least moving this patch up before patch 14.

author maintainer
#76 lore raw

PATCH v2 24/30 vfio-pci/zdev: wire up group notifier

In reply to Matthew Rosato

KVM zPCI passthrough device logic will need a reference to the associated kvm guest that has access to the device. Let's register a group notifier for VFIO_GROUP_NOTIFY_SET_KVM to catch this information in order to create an association between a kvm guest and the host zdev.

---
 arch/s390/include/asm/kvm_pci.h  |  2 ++
 drivers/vfio/pci/vfio_pci_core.c |  2 ++
 drivers/vfio/pci/vfio_pci_zdev.c | 46 ++++++++++++++++++++++++++++++++
 include/linux/vfio_pci_core.h    | 10 +++++++
 4 files changed, 60 insertions(+)
arch/s390/include/asm/kvm_pci.h +2 -0
index fa90729a35cf..97a90b37c87d 100644
@@ -17,6 +17,7 @@
 #include <linux/kvm.h>
 #include <linux/pci.h>
 #include <linux/mutex.h>
+#include <linux/notifier.h>
 #include <asm/pci_insn.h>
 #include <asm/pci_dma.h>
 
@@ -33,6 +34,7 @@ struct kvm_zdev {
 	u64 rpcit_count;
 	struct kvm_zdev_ioat ioat;
 	struct zpci_fib fib;
+	struct notifier_block nb;
 };
 
 int kvm_s390_pci_dev_open(struct zpci_dev *zdev);
drivers/vfio/pci/vfio_pci_core.c +2 -0
index f948e6cd2993..fc57d4d0abbe 100644
@@ -452,6 +452,7 @@ void vfio_pci_core_close_device(struct vfio_device *core_vdev)
 
 	vfio_pci_vf_token_user_add(vdev, -1);
 	vfio_spapr_pci_eeh_release(vdev->pdev);
+	vfio_pci_zdev_release(vdev);
 	vfio_pci_core_disable(vdev);
 
 	mutex_lock(&vdev->igate);
@@ -470,6 +471,7 @@ EXPORT_SYMBOL_GPL(vfio_pci_core_close_device);
 void vfio_pci_core_finish_enable(struct vfio_pci_core_device *vdev)
 {
 	vfio_pci_probe_mmaps(vdev);
+	vfio_pci_zdev_open(vdev);
 	vfio_spapr_pci_eeh_open(vdev->pdev);
 	vfio_pci_vf_token_user_add(vdev, 1);
 }
drivers/vfio/pci/vfio_pci_zdev.c +46 -0

55 lines not on this page · raw message

include/linux/vfio_pci_core.h +10 -0
index 5e2bca3b89db..05287f8ac855 100644
@@ -198,12 +198,22 @@ static inline int vfio_pci_igd_init(struct vfio_pci_core_device *vdev)
 #ifdef CONFIG_VFIO_PCI_ZDEV
 extern int vfio_pci_info_zdev_add_caps(struct vfio_pci_core_device *vdev,
 				       struct vfio_info_cap *caps);
+void vfio_pci_zdev_open(struct vfio_pci_core_device *vdev);
+void vfio_pci_zdev_release(struct vfio_pci_core_device *vdev);
 #else
 static inline int vfio_pci_info_zdev_add_caps(struct vfio_pci_core_device *vdev,
 					      struct vfio_info_cap *caps)
 {
 	return -ENODEV;
 }
+
+static inline void vfio_pci_zdev_open(struct vfio_pci_core_device *vdev)
+{
+}
+
+static inline void vfio_pci_zdev_release(struct vfio_pci_core_device *vdev)
+{
+}
 #endif
 
 /* Will be exported for vfio pci drivers usage */
Signature · 1 line
2.27.0
#77 lore raw

Re: [PATCH v2 24/30] vfio-pci/zdev: wire up group notifier

Quoted 69 lines · Matthew Rosato

KVM zPCI passthrough device logic will need a reference to the associated kvm guest that has access to the device. Let's register a group notifier for VFIO_GROUP_NOTIFY_SET_KVM to catch this information in order to create an association between a kvm guest and the host zdev.

---
  arch/s390/include/asm/kvm_pci.h  |  2 ++
  drivers/vfio/pci/vfio_pci_core.c |  2 ++
  drivers/vfio/pci/vfio_pci_zdev.c | 46 ++++++++++++++++++++++++++++++++
  include/linux/vfio_pci_core.h    | 10 +++++++
  4 files changed, 60 insertions(+)
arch/s390/include/asm/kvm_pci.h +2 -0
index fa90729a35cf..97a90b37c87d 100644
@@ -17,6 +17,7 @@
  #include <linux/kvm.h>
  #include <linux/pci.h>
  #include <linux/mutex.h>
+#include <linux/notifier.h>
  #include <asm/pci_insn.h>
  #include <asm/pci_dma.h>
 
@@ -33,6 +34,7 @@ struct kvm_zdev {
  	u64 rpcit_count;
  	struct kvm_zdev_ioat ioat;
  	struct zpci_fib fib;
+	struct notifier_block nb;
  };
 
  int kvm_s390_pci_dev_open(struct zpci_dev *zdev);
drivers/vfio/pci/vfio_pci_core.c +2 -0
index f948e6cd2993..fc57d4d0abbe 100644
@@ -452,6 +452,7 @@ void vfio_pci_core_close_device(struct vfio_device *core_vdev)
 
  	vfio_pci_vf_token_user_add(vdev, -1);
  	vfio_spapr_pci_eeh_release(vdev->pdev);
+	vfio_pci_zdev_release(vdev);
  	vfio_pci_core_disable(vdev);
 
  	mutex_lock(&vdev->igate);
@@ -470,6 +471,7 @@ EXPORT_SYMBOL_GPL(vfio_pci_core_close_device);
  void vfio_pci_core_finish_enable(struct vfio_pci_core_device *vdev)
  {
  	vfio_pci_probe_mmaps(vdev);
+	vfio_pci_zdev_open(vdev);
  	vfio_spapr_pci_eeh_open(vdev->pdev);
  	vfio_pci_vf_token_user_add(vdev, 1);
  }
drivers/vfio/pci/vfio_pci_zdev.c +1 -0
index ea4c0d2b0663..5c2bddc57b39 100644
@@ -13,6 +13,7 @@
  #include <linux/vfio_zdev.h>
  #include <asm/pci_clp.h>
  #include <asm/pci_io.h>
+#include <asm/kvm_pci.h>
 
  #include <linux/vfio_pci_core.h>
 
@@ -136,3 +137,48 @@ int vfio_pci_info_zdev_add_caps(struct vfio_pci_core_device *vdev,
 
  	return ret;
  }

drivers/vfio/pci/vfio_pci_zdev.c int vfio_pci_info_zdev_add_caps(struct vfio_pci_core_device *vdev, · line 149

Quoted 10 lines
drivers/vfio/pci/vfio_pci_zdev.c +10 -0
+
+static int vfio_pci_zdev_group_notifier(struct notifier_block *nb,
+					unsigned long action, void *data)
+{
+	struct kvm_zdev *kzdev = container_of(nb, struct kvm_zdev, nb);
+
+	if (action == VFIO_GROUP_NOTIFY_SET_KVM) {
+		if (!data || !kzdev->zdev)
+			return NOTIFY_DONE;
+		kvm_s390_pci_attach_kvm(kzdev->zdev, data);

Why not just set kzdev->kvm = data ?

alternatively, define kvm_s390_pci_attach_kvm() as an inline instead of a global function.

otherwise LGTM

Quoted 63 lines
drivers/vfio/pci/vfio_pci_zdev.c +35 -0
+	}
+
+	return NOTIFY_OK;
+}
+
+void vfio_pci_zdev_open(struct vfio_pci_core_device *vdev)
+{
+	unsigned long events = VFIO_GROUP_NOTIFY_SET_KVM;
+	struct zpci_dev *zdev = to_zpci(vdev->pdev);
+
+	if (!zdev)
+		return;
+
+	if (kvm_s390_pci_dev_open(zdev))
+		return;
+
+	zdev->kzdev->nb.notifier_call = vfio_pci_zdev_group_notifier;
+
+	if (vfio_register_notifier(vdev->vdev.dev, VFIO_GROUP_NOTIFY,
+				   &events, &zdev->kzdev->nb))
+		kvm_s390_pci_dev_release(zdev);
+}
+
+void vfio_pci_zdev_release(struct vfio_pci_core_device *vdev)
+{
+	struct zpci_dev *zdev = to_zpci(vdev->pdev);
+
+	if (!zdev || !zdev->kzdev)
+		return;
+
+	vfio_unregister_notifier(vdev->vdev.dev, VFIO_GROUP_NOTIFY,
+				 &zdev->kzdev->nb);
+
+	kvm_s390_pci_dev_release(zdev);
+}
include/linux/vfio_pci_core.h +10 -0
index 5e2bca3b89db..05287f8ac855 100644
@@ -198,12 +198,22 @@ static inline int vfio_pci_igd_init(struct vfio_pci_core_device *vdev)
  #ifdef CONFIG_VFIO_PCI_ZDEV
  extern int vfio_pci_info_zdev_add_caps(struct vfio_pci_core_device *vdev,
  				       struct vfio_info_cap *caps);
+void vfio_pci_zdev_open(struct vfio_pci_core_device *vdev);
+void vfio_pci_zdev_release(struct vfio_pci_core_device *vdev);
  #else
  static inline int vfio_pci_info_zdev_add_caps(struct vfio_pci_core_device *vdev,
  					      struct vfio_info_cap *caps)
  {
  	return -ENODEV;
  }
+
+static inline void vfio_pci_zdev_open(struct vfio_pci_core_device *vdev)
+{
+}
+
+static inline void vfio_pci_zdev_release(struct vfio_pci_core_device *vdev)
+{
+}
  #endif
 
  /* Will be exported for vfio pci drivers usage */
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#78 lore raw

Re: [PATCH v2 24/30] vfio-pci/zdev: wire up group notifier

Quoted 80 lines · Pierre Morel
Quoted 64 lines · Matthew Rosato

KVM zPCI passthrough device logic will need a reference to the associated kvm guest that has access to the device.  Let's register a group notifier for VFIO_GROUP_NOTIFY_SET_KVM to catch this information in order to create an association between a kvm guest and the host zdev.

---
  arch/s390/include/asm/kvm_pci.h  |  2 ++ drivers/vfio/pci/vfio_pci_core.c |  2 ++ drivers/vfio/pci/vfio_pci_zdev.c | 46 ++++++++++++++++++++++++++++++++ include/linux/vfio_pci_core.h    | 10 +++++++

  4 files changed, 60 insertions(+)
arch/s390/include/asm/kvm_pci.h +2 -0
index fa90729a35cf..97a90b37c87d 100644
@@ -17,6 +17,7 @@
   #include <linux/kvm.h>
   #include <linux/pci.h>
   #include <linux/mutex.h>
+#include <linux/notifier.h>
   #include <asm/pci_insn.h>
   #include <asm/pci_dma.h>
@@ -33,6 +34,7 @@ struct kvm_zdev {
       u64 rpcit_count;
       struct kvm_zdev_ioat ioat;
       struct zpci_fib fib;
+    struct notifier_block nb;
   };
   int kvm_s390_pci_dev_open(struct zpci_dev *zdev);
drivers/vfio/pci/vfio_pci_core.c +2 -0
index f948e6cd2993..fc57d4d0abbe 100644
@@ -452,6 +452,7 @@ void vfio_pci_core_close_device(struct vfio_device 
 *core_vdev)
       vfio_pci_vf_token_user_add(vdev, -1);
       vfio_spapr_pci_eeh_release(vdev->pdev);
+    vfio_pci_zdev_release(vdev);
       vfio_pci_core_disable(vdev);
       mutex_lock(&vdev->igate);
@@ -470,6 +471,7 @@ EXPORT_SYMBOL_GPL(vfio_pci_core_close_device);
   void vfio_pci_core_finish_enable(struct vfio_pci_core_device *vdev)
   {
       vfio_pci_probe_mmaps(vdev);
+    vfio_pci_zdev_open(vdev);
       vfio_spapr_pci_eeh_open(vdev->pdev);
       vfio_pci_vf_token_user_add(vdev, 1);
   }
drivers/vfio/pci/vfio_pci_zdev.c +1 -0
index ea4c0d2b0663..5c2bddc57b39 100644
@@ -13,6 +13,7 @@
   #include <linux/vfio_zdev.h>
   #include <asm/pci_clp.h>
   #include <asm/pci_io.h>
+#include <asm/kvm_pci.h>
   #include <linux/vfio_pci_core.h>
@@ -136,3 +137,48 @@ int vfio_pci_info_zdev_add_caps(struct 
 vfio_pci_core_device *vdev,
       return ret;
   }

drivers/vfio/pci/vfio_pci_zdev.c int vfio_pci_info_zdev_add_caps(struct · line 149

Quoted 10 lines
drivers/vfio/pci/vfio_pci_zdev.c +10 -0
+
+static int vfio_pci_zdev_group_notifier(struct notifier_block *nb,
+                    unsigned long action, void *data)
+{
+    struct kvm_zdev *kzdev = container_of(nb, struct kvm_zdev, nb);
+
+    if (action == VFIO_GROUP_NOTIFY_SET_KVM) {
+        if (!data || !kzdev->zdev)
+            return NOTIFY_DONE;
+        kvm_s390_pci_attach_kvm(kzdev->zdev, data);

Why not just set kzdev->kvm = data ?

Quoted 3 lines

alternatively, define kvm_s390_pci_attach_kvm() as an inline instead of a global function.

otherwise LGTM

At some point in the past this function did more than just set a pointer... You are correct there's no need for this abstraction now, let's just set kzdev->kvm = data directly here and drop the kvm_s390_pci_attach_kvm function.

author maintainer
#79 lore raw

PATCH v2 25/30 vfio-pci/zdev: wire up zPCI interpretive execution support

In reply to Matthew Rosato

Introduce support for VFIO_DEVICE_FEATURE_ZPCI_INTERP, which is a new VFIO_DEVICE_FEATURE ioctl. This interface is used to indicate that an s390x vfio-pci device wishes to enable/disable zPCI interpretive execution, which allows zPCI instructions to be executed directly by underlying firmware without KVM involvement.

---
 arch/s390/include/asm/kvm_pci.h  |  1 +
 drivers/vfio/pci/vfio_pci_core.c |  2 +
 drivers/vfio/pci/vfio_pci_zdev.c | 78 ++++++++++++++++++++++++++++++++
 include/linux/vfio_pci_core.h    | 10 ++++
 include/uapi/linux/vfio.h        |  7 +++
 include/uapi/linux/vfio_zdev.h   | 15 ++++++
 6 files changed, 113 insertions(+)
arch/s390/include/asm/kvm_pci.h +1 -0
index 97a90b37c87d..dc00c3f27a00 100644
@@ -35,6 +35,7 @@ struct kvm_zdev {
 	struct kvm_zdev_ioat ioat;
 	struct zpci_fib fib;
 	struct notifier_block nb;
+	bool interp;
 };
 
 int kvm_s390_pci_dev_open(struct zpci_dev *zdev);
drivers/vfio/pci/vfio_pci_core.c +2 -0
index fc57d4d0abbe..2b2d64a2190c 100644
@@ -1172,6 +1172,8 @@ long vfio_pci_core_ioctl(struct vfio_device *core_vdev, unsigned int cmd,
 			mutex_unlock(&vdev->vf_token->lock);
 
 			return 0;
+		case VFIO_DEVICE_FEATURE_ZPCI_INTERP:
+			return vfio_pci_zdev_feat_interp(vdev, feature, arg);
 		default:
 			return -ENOTTY;
 		}
drivers/vfio/pci/vfio_pci_zdev.c +78 -0

101 lines not on this page · raw message

include/linux/vfio_pci_core.h +10 -0
index 05287f8ac855..0db2b1051931 100644
@@ -198,6 +198,9 @@ static inline int vfio_pci_igd_init(struct vfio_pci_core_device *vdev)
 #ifdef CONFIG_VFIO_PCI_ZDEV
 extern int vfio_pci_info_zdev_add_caps(struct vfio_pci_core_device *vdev,
 				       struct vfio_info_cap *caps);
+int vfio_pci_zdev_feat_interp(struct vfio_pci_core_device *vdev,
+			      struct vfio_device_feature feature,
+			      unsigned long arg);
 void vfio_pci_zdev_open(struct vfio_pci_core_device *vdev);
 void vfio_pci_zdev_release(struct vfio_pci_core_device *vdev);
 #else
@@ -207,6 +210,13 @@ static inline int vfio_pci_info_zdev_add_caps(struct vfio_pci_core_device *vdev,
 	return -ENODEV;
 }
 
+static inline int vfio_pci_zdev_feat_interp(struct vfio_pci_core_device *vdev,
+					    struct vfio_device_feature feature,
+					    unsigned long arg)
+{
+	return -ENOTTY;
+}
+
 static inline void vfio_pci_zdev_open(struct vfio_pci_core_device *vdev)
 {
 }
include/uapi/linux/vfio.h +7 -0
index ef33ea002b0b..b9a75485b8e7 100644
@@ -1002,6 +1002,13 @@ struct vfio_device_feature {
  */
 #define VFIO_DEVICE_FEATURE_PCI_VF_TOKEN	(0)
 
+/*
+ * Provide support for enabling interpretation of zPCI instructions.  This
+ * feature is only valid for s390x PCI devices.  Data provided when setting
+ * and getting this feature is futher described in vfio_zdev.h
+ */
+#define VFIO_DEVICE_FEATURE_ZPCI_INTERP		(1)
+
 /* -------- API for Type1 VFIO IOMMU -------- */
 
 /**
include/uapi/linux/vfio_zdev.h +15 -0
index b4309397b6b2..575f0410dc66 100644
@@ -75,4 +75,19 @@ struct vfio_device_info_cap_zpci_pfip {
 	__u8 pfip[];
 };
 
+/**
+ * VFIO_DEVICE_FEATURE_ZPCI_INTERP
+ *
+ * This feature is used for enabling zPCI instruction interpretation for a
+ * device.  No data is provided when setting this feature.  When getting
+ * this feature, the following structure is provided which details whether
+ * or not interpretation is active and provides the guest with host device
+ * information necessary to enable interpretation.
+ */
+struct vfio_device_zpci_interp {
+	__u64 flags;
+#define VFIO_DEVICE_ZPCI_FLAG_INTERP 1
+	__u32 fh;		/* Host device function handle */
+};
+
 #endif
Signature · 1 line
2.27.0
#80 lore raw

Re: [PATCH v2 25/30] vfio-pci/zdev: wire up zPCI interpretive execution support

Quoted 21 lines · Matthew Rosato

Introduce support for VFIO_DEVICE_FEATURE_ZPCI_INTERP, which is a new VFIO_DEVICE_FEATURE ioctl. This interface is used to indicate that an s390x vfio-pci device wishes to enable/disable zPCI interpretive execution, which allows zPCI instructions to be executed directly by underlying firmware without KVM involvement.

---
  arch/s390/include/asm/kvm_pci.h  |  1 +
  drivers/vfio/pci/vfio_pci_core.c |  2 +
  drivers/vfio/pci/vfio_pci_zdev.c | 78 ++++++++++++++++++++++++++++++++
  include/linux/vfio_pci_core.h    | 10 ++++
  include/uapi/linux/vfio.h        |  7 +++
  include/uapi/linux/vfio_zdev.h   | 15 ++++++
  6 files changed, 113 insertions(+)
arch/s390/include/asm/kvm_pci.h +0 -0
index 97a90b37c87d..dc00c3f27a00 100644

arch/s390/include/asm/kvm_pci.h struct kvm_zdev { · line 38

Quoted 4 lines
arch/s390/include/asm/kvm_pci.h +1 -0
@@ -35,6 +35,7 @@ struct kvm_zdev {
  	struct kvm_zdev_ioat ioat;
  	struct zpci_fib fib;
  	struct notifier_block nb;
+	bool interp;

NIT: s/interp/interpretation/ ?

Quoted 46 lines
arch/s390/include/asm/kvm_pci.h +0 -0
  };
 
  int kvm_s390_pci_dev_open(struct zpci_dev *zdev);
drivers/vfio/pci/vfio_pci_core.c +2 -0
index fc57d4d0abbe..2b2d64a2190c 100644
@@ -1172,6 +1172,8 @@ long vfio_pci_core_ioctl(struct vfio_device *core_vdev, unsigned int cmd,
  			mutex_unlock(&vdev->vf_token->lock);
 
  			return 0;
+		case VFIO_DEVICE_FEATURE_ZPCI_INTERP:
+			return vfio_pci_zdev_feat_interp(vdev, feature, arg);
  		default:
  			return -ENOTTY;
  		}
drivers/vfio/pci/vfio_pci_zdev.c +15 -0
index 5c2bddc57b39..4339f48b98bc 100644
@@ -54,6 +54,10 @@ static int zpci_group_cap(struct zpci_dev *zdev, struct vfio_info_cap *caps)
  		.version = zdev->version
  	};
 
+	/* Some values are different for interpreted devices */
+	if (zdev->kzdev && zdev->kzdev->interp)
+		cap.maxstbl = zdev->maxstbl;
+
  	return vfio_info_add_capability(caps, &cap.header, sizeof(cap));
  }
 
@@ -138,6 +142,72 @@ int vfio_pci_info_zdev_add_caps(struct vfio_pci_core_device *vdev,
  	return ret;
  }
 
+int vfio_pci_zdev_feat_interp(struct vfio_pci_core_device *vdev,
+			      struct vfio_device_feature feature,
+			      unsigned long arg)
+{
+	struct zpci_dev *zdev = to_zpci(vdev->pdev);
+	struct vfio_device_zpci_interp *data;
+	struct vfio_device_feature *feat;
+	unsigned long minsz;
+	int size, rc;
+
+	if (!zdev || !zdev->kzdev)

drivers/vfio/pci/vfio_pci_zdev.c int vfio_pci_info_zdev_add_caps(struct vfio_pci_core_device *vdev, · line 165

Quoted 10 lines
drivers/vfio/pci/vfio_pci_zdev.c +10 -0
+		return -EINVAL;
+
+	/* If PROBE specified, return probe results immediately */
+	if (feature.flags & VFIO_DEVICE_FEATURE_PROBE)
+		return kvm_s390_pci_interp_probe(zdev);
+
+	/* GET and SET are mutually exclusive */
+	if ((feature.flags & VFIO_DEVICE_FEATURE_GET) &&
+	    (feature.flags & VFIO_DEVICE_FEATURE_SET))
+		return -EINVAL;

Isn't the check already done in VFIO core?

Quoted 130 lines
drivers/vfio/pci/vfio_pci_zdev.c +53 -0
+
+	size = sizeof(*feat) + sizeof(*data);
+	feat = kzalloc(size, GFP_KERNEL);
+	if (!feat)
+		return -ENOMEM;
+
+	data = (struct vfio_device_zpci_interp *)&feat->data;
+	minsz = offsetofend(struct vfio_device_feature, flags);
+
+	if (feature.argsz < minsz + sizeof(*data))
+		return -EINVAL;
+
+	/* Get the rest of the payload for GET/SET */
+	rc = copy_from_user(data, (void __user *)(arg + minsz),
+			    sizeof(*data));
+	if (rc)
+		rc = -EINVAL;
+
+	if (feature.flags & VFIO_DEVICE_FEATURE_GET) {
+		if (zdev->gd != 0)
+			data->flags = VFIO_DEVICE_ZPCI_FLAG_INTERP;
+		else
+			data->flags = 0;
+		data->fh = zdev->fh;
+		/* userspace is using host fh, give interpreted clp values */
+		zdev->kzdev->interp = true;
+
+		if (copy_to_user((void __user *)arg, feat, size))
+			rc = -EFAULT;
+	} else if (feature.flags & VFIO_DEVICE_FEATURE_SET) {
+		if (data->flags == VFIO_DEVICE_ZPCI_FLAG_INTERP)
+			rc = kvm_s390_pci_interp_enable(zdev);
+		else if (data->flags == 0)
+			rc = kvm_s390_pci_interp_disable(zdev);
+		else
+			rc = -EINVAL;
+	} else {
+		/* Neither GET nor SET were specified */
+		rc = -EINVAL;
+	}
+
+	kfree(feat);
+	return rc;
+}
+
  static int vfio_pci_zdev_group_notifier(struct notifier_block *nb,
  					unsigned long action, void *data)
  {
@@ -164,6 +234,7 @@ void vfio_pci_zdev_open(struct vfio_pci_core_device *vdev)
  		return;
 
  	zdev->kzdev->nb.notifier_call = vfio_pci_zdev_group_notifier;
+	zdev->kzdev->interp = false;
 
  	if (vfio_register_notifier(vdev->vdev.dev, VFIO_GROUP_NOTIFY,
  				   &events, &zdev->kzdev->nb))
@@ -180,5 +251,12 @@ void vfio_pci_zdev_release(struct vfio_pci_core_device *vdev)
  	vfio_unregister_notifier(vdev->vdev.dev, VFIO_GROUP_NOTIFY,
  				 &zdev->kzdev->nb);
 
+	/*
+	 * If the device was using interpretation, don't trust that userspace
+	 * did the appropriate cleanup
+	 */
+	if (zdev->gd != 0)
+		kvm_s390_pci_interp_disable(zdev);
+
  	kvm_s390_pci_dev_release(zdev);
  }
include/linux/vfio_pci_core.h +10 -0
index 05287f8ac855..0db2b1051931 100644
@@ -198,6 +198,9 @@ static inline int vfio_pci_igd_init(struct vfio_pci_core_device *vdev)
  #ifdef CONFIG_VFIO_PCI_ZDEV
  extern int vfio_pci_info_zdev_add_caps(struct vfio_pci_core_device *vdev,
  				       struct vfio_info_cap *caps);
+int vfio_pci_zdev_feat_interp(struct vfio_pci_core_device *vdev,
+			      struct vfio_device_feature feature,
+			      unsigned long arg);
  void vfio_pci_zdev_open(struct vfio_pci_core_device *vdev);
  void vfio_pci_zdev_release(struct vfio_pci_core_device *vdev);
  #else
@@ -207,6 +210,13 @@ static inline int vfio_pci_info_zdev_add_caps(struct vfio_pci_core_device *vdev,
  	return -ENODEV;
  }
 
+static inline int vfio_pci_zdev_feat_interp(struct vfio_pci_core_device *vdev,
+					    struct vfio_device_feature feature,
+					    unsigned long arg)
+{
+	return -ENOTTY;
+}
+
  static inline void vfio_pci_zdev_open(struct vfio_pci_core_device *vdev)
  {
  }
include/uapi/linux/vfio.h +7 -0
index ef33ea002b0b..b9a75485b8e7 100644
@@ -1002,6 +1002,13 @@ struct vfio_device_feature {
   */
  #define VFIO_DEVICE_FEATURE_PCI_VF_TOKEN	(0)
 
+/*
+ * Provide support for enabling interpretation of zPCI instructions.  This
+ * feature is only valid for s390x PCI devices.  Data provided when setting
+ * and getting this feature is futher described in vfio_zdev.h
+ */
+#define VFIO_DEVICE_FEATURE_ZPCI_INTERP		(1)
+
  /* -------- API for Type1 VFIO IOMMU -------- */
 
  /**
include/uapi/linux/vfio_zdev.h +6 -0
index b4309397b6b2..575f0410dc66 100644
@@ -75,4 +75,19 @@ struct vfio_device_info_cap_zpci_pfip {
  	__u8 pfip[];
  };
 
+/**
+ * VFIO_DEVICE_FEATURE_ZPCI_INTERP
+ *
+ * This feature is used for enabling zPCI instruction interpretation for a
+ * device.  No data is provided when setting this feature.  When getting
+ * this feature, the following structure is provided which details whether

include/uapi/linux/vfio_zdev.h struct vfio_device_info_cap_zpci_pfip { · line 93

Quoted 10 lines
include/uapi/linux/vfio_zdev.h +9 -0
+ * or not interpretation is active and provides the guest with host device
+ * information necessary to enable interpretation.
+ */
+struct vfio_device_zpci_interp {
+	__u64 flags;
+#define VFIO_DEVICE_ZPCI_FLAG_INTERP 1
+	__u32 fh;		/* Host device function handle */
+};
+
  #endif

Otherwise LGTM

Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#81 lore raw

Re: [PATCH v2 25/30] vfio-pci/zdev: wire up zPCI interpretive execution support

Quoted 19 lines · Pierre Morel
Quoted 16 lines · Matthew Rosato

Introduce support for VFIO_DEVICE_FEATURE_ZPCI_INTERP, which is a new VFIO_DEVICE_FEATURE ioctl.  This interface is used to indicate that an s390x vfio-pci device wishes to enable/disable zPCI interpretive execution, which allows zPCI instructions to be executed directly by underlying firmware without KVM involvement.

---
  arch/s390/include/asm/kvm_pci.h  |  1 + drivers/vfio/pci/vfio_pci_core.c |  2 + drivers/vfio/pci/vfio_pci_zdev.c | 78 ++++++++++++++++++++++++++++++++ include/linux/vfio_pci_core.h    | 10 ++++ include/uapi/linux/vfio.h        |  7 +++ include/uapi/linux/vfio_zdev.h   | 15 ++++++

  6 files changed, 113 insertions(+)
Quoted 11 lines

arch/s390/include/asm/kvm_pci.h struct kvm_zdev { · line 38

Quoted 9 lines
arch/s390/include/asm/kvm_pci.h +1 -0
index 97a90b37c87d..dc00c3f27a00 100644
@@ -35,6 +35,7 @@ struct kvm_zdev {
       struct kvm_zdev_ioat ioat;
       struct zpci_fib fib;
       struct notifier_block nb;
+    bool interp;

NIT: s/interp/interpretation/ ?

OK

Quoted 49 lines
Quoted 49 lines
arch/s390/include/asm/kvm_pci.h +0 -0
   };
   int kvm_s390_pci_dev_open(struct zpci_dev *zdev);
drivers/vfio/pci/vfio_pci_core.c +2 -0
index fc57d4d0abbe..2b2d64a2190c 100644
@@ -1172,6 +1172,8 @@ long vfio_pci_core_ioctl(struct vfio_device 
 *core_vdev, unsigned int cmd,
               mutex_unlock(&vdev->vf_token->lock);
               return 0;
+        case VFIO_DEVICE_FEATURE_ZPCI_INTERP:
+            return vfio_pci_zdev_feat_interp(vdev, feature, arg);

          default:
              return -ENOTTY;
          }

drivers/vfio/pci/vfio_pci_zdev.c +0 -0
index 5c2bddc57b39..4339f48b98bc 100644
@@ -54,6 +54,10 @@ static int zpci_group_cap(struct zpci_dev *zdev, 
 struct vfio_info_cap *caps)

          .version = zdev->version
      };
+    /* Some values are different for interpreted devices */ +    if (zdev->kzdev && zdev->kzdev->interp)
+        cap.maxstbl = zdev->maxstbl;
+
      return vfio_info_add_capability(caps, &cap.header, sizeof(cap)); }

drivers/vfio/pci/vfio_pci_zdev.c +15 -0
@@ -138,6 +142,72 @@ int vfio_pci_info_zdev_add_caps(struct 
 vfio_pci_core_device *vdev,
       return ret;
   }
+int vfio_pci_zdev_feat_interp(struct vfio_pci_core_device *vdev,
+                  struct vfio_device_feature feature,
+                  unsigned long arg)
+{
+    struct zpci_dev *zdev = to_zpci(vdev->pdev);
+    struct vfio_device_zpci_interp *data;
+    struct vfio_device_feature *feat;
+    unsigned long minsz;
+    int size, rc;
+
+    if (!zdev || !zdev->kzdev)
+        return -EINVAL;
+
+    /* If PROBE specified, return probe results immediately */
+    if (feature.flags & VFIO_DEVICE_FEATURE_PROBE)
Quoted 8 lines

drivers/vfio/pci/vfio_pci_zdev.c int vfio_pci_info_zdev_add_caps(struct vfio_pci_core_device *vdev, · line 165

Quoted 6 lines
drivers/vfio/pci/vfio_pci_zdev.c +6 -0
+        return kvm_s390_pci_interp_probe(zdev);
+
+    /* GET and SET are mutually exclusive */
+    if ((feature.flags & VFIO_DEVICE_FEATURE_GET) &&
+        (feature.flags & VFIO_DEVICE_FEATURE_SET))
+        return -EINVAL;

Isn't the check already done in VFIO core?

Oh, yes you are correct. Then this can be removed for this patch as well as the next 2 patches.

author maintainer
#82 lore raw

PATCH v2 26/30 vfio-pci/zdev: wire up zPCI adapter interrupt forwarding support

In reply to Matthew Rosato

Introduce support for VFIO_DEVICE_FEATURE_ZPCI_AIF, which is a new VFIO_DEVICE_FEATURE ioctl. This interface is used to indicate that an s390x vfio-pci device wishes to enable/disable zPCI adapter interrupt forwarding, which allows underlying firmware to deliver interrupts directly to the associated kvm guest.

---
 arch/s390/include/asm/kvm_pci.h  |  2 +
 drivers/vfio/pci/vfio_pci_core.c |  2 +
 drivers/vfio/pci/vfio_pci_zdev.c | 98 +++++++++++++++++++++++++++++++-
 include/linux/vfio_pci_core.h    | 10 ++++
 include/uapi/linux/vfio.h        |  7 +++
 include/uapi/linux/vfio_zdev.h   | 20 +++++++
 6 files changed, 138 insertions(+), 1 deletion(-)
arch/s390/include/asm/kvm_pci.h +2 -0
index dc00c3f27a00..dbab349a4a75 100644
@@ -36,6 +36,8 @@ struct kvm_zdev {
 	struct zpci_fib fib;
 	struct notifier_block nb;
 	bool interp;
+	bool aif;
+	bool fhost;
 };
 
 int kvm_s390_pci_dev_open(struct zpci_dev *zdev);
drivers/vfio/pci/vfio_pci_core.c +2 -0
index 2b2d64a2190c..01658de660bd 100644
@@ -1174,6 +1174,8 @@ long vfio_pci_core_ioctl(struct vfio_device *core_vdev, unsigned int cmd,
 			return 0;
 		case VFIO_DEVICE_FEATURE_ZPCI_INTERP:
 			return vfio_pci_zdev_feat_interp(vdev, feature, arg);
+		case VFIO_DEVICE_FEATURE_ZPCI_AIF:
+			return vfio_pci_zdev_feat_aif(vdev, feature, arg);
 		default:
 			return -ENOTTY;
 		}
drivers/vfio/pci/vfio_pci_zdev.c +97 -1

117 lines not on this page · raw message

include/linux/vfio_pci_core.h +10 -0
index 0db2b1051931..7ec5e82e7933 100644
@@ -201,6 +201,9 @@ extern int vfio_pci_info_zdev_add_caps(struct vfio_pci_core_device *vdev,
 int vfio_pci_zdev_feat_interp(struct vfio_pci_core_device *vdev,
 			      struct vfio_device_feature feature,
 			      unsigned long arg);
+int vfio_pci_zdev_feat_aif(struct vfio_pci_core_device *vdev,
+			   struct vfio_device_feature feature,
+			   unsigned long arg);
 void vfio_pci_zdev_open(struct vfio_pci_core_device *vdev);
 void vfio_pci_zdev_release(struct vfio_pci_core_device *vdev);
 #else
@@ -217,6 +220,13 @@ static inline int vfio_pci_zdev_feat_interp(struct vfio_pci_core_device *vdev,
 	return -ENOTTY;
 }
 
+static inline int vfio_pci_zdev_feat_aif(struct vfio_pci_core_device *vdev,
+					 struct vfio_device_feature feature,
+					 unsigned long arg)
+{
+	return -ENOTTY;
+}
+
 static inline void vfio_pci_zdev_open(struct vfio_pci_core_device *vdev)
 {
 }
include/uapi/linux/vfio.h +7 -0
index b9a75485b8e7..fe3bfd99bf50 100644
@@ -1009,6 +1009,13 @@ struct vfio_device_feature {
  */
 #define VFIO_DEVICE_FEATURE_ZPCI_INTERP		(1)
 
+/*
+ * Provide support for enbaling adapter interruption forwarding for zPCI
+ * devices.  This feature is only valid for s390x PCI devices.  Data provided
+ * when setting and getting this feature is further described in vfio_zdev.h
+ */
+#define VFIO_DEVICE_FEATURE_ZPCI_AIF		(2)
+
 /* -------- API for Type1 VFIO IOMMU -------- */
 
 /**
include/uapi/linux/vfio_zdev.h +20 -0
index 575f0410dc66..c574e23f9385 100644
@@ -90,4 +90,24 @@ struct vfio_device_zpci_interp {
 	__u32 fh;		/* Host device function handle */
 };
 
+/**
+ * VFIO_DEVICE_FEATURE_ZPCI_AIF
+ *
+ * This feature is used for enabling forwarding of adapter interrupts directly
+ * from firmware to the guest.  When setting this feature, the flags indicate
+ * whether to enable/disable the feature and the structure defined below is
+ * used to setup the forwarding structures.  When getting this feature, only
+ * the flags are used to indicate the current state.
+ */
+struct vfio_device_zpci_aif {
+	__u64 flags;
+#define VFIO_DEVICE_ZPCI_FLAG_AIF_FLOAT 1
+#define VFIO_DEVICE_ZPCI_FLAG_AIF_HOST 2
+	__u64 ibv;		/* Address of guest interrupt bit vector */
+	__u64 sb;		/* Address of guest summary bit */
+	__u32 noi;		/* Number of interrupts */
+	__u8 isc;		/* Guest interrupt subclass */
+	__u8 sbo;		/* Offset of guest summary bit vector */
+};
+
 #endif
Signature · 1 line
2.27.0
#83 lore raw

Re: [PATCH v2 26/30] vfio-pci/zdev: wire up zPCI adapter interrupt forwarding support

Quoted 223 lines · Matthew Rosato

Introduce support for VFIO_DEVICE_FEATURE_ZPCI_AIF, which is a new VFIO_DEVICE_FEATURE ioctl. This interface is used to indicate that an s390x vfio-pci device wishes to enable/disable zPCI adapter interrupt forwarding, which allows underlying firmware to deliver interrupts directly to the associated kvm guest.

---
  arch/s390/include/asm/kvm_pci.h  |  2 +
  drivers/vfio/pci/vfio_pci_core.c |  2 +
  drivers/vfio/pci/vfio_pci_zdev.c | 98 +++++++++++++++++++++++++++++++-
  include/linux/vfio_pci_core.h    | 10 ++++
  include/uapi/linux/vfio.h        |  7 +++
  include/uapi/linux/vfio_zdev.h   | 20 +++++++
  6 files changed, 138 insertions(+), 1 deletion(-)
arch/s390/include/asm/kvm_pci.h +2 -0
index dc00c3f27a00..dbab349a4a75 100644
@@ -36,6 +36,8 @@ struct kvm_zdev {
  	struct zpci_fib fib;
  	struct notifier_block nb;
  	bool interp;
+	bool aif;
+	bool fhost;
  };
 
  int kvm_s390_pci_dev_open(struct zpci_dev *zdev);
drivers/vfio/pci/vfio_pci_core.c +2 -0
index 2b2d64a2190c..01658de660bd 100644
@@ -1174,6 +1174,8 @@ long vfio_pci_core_ioctl(struct vfio_device *core_vdev, unsigned int cmd,
  			return 0;
  		case VFIO_DEVICE_FEATURE_ZPCI_INTERP:
  			return vfio_pci_zdev_feat_interp(vdev, feature, arg);
+		case VFIO_DEVICE_FEATURE_ZPCI_AIF:
+			return vfio_pci_zdev_feat_aif(vdev, feature, arg);
  		default:
  			return -ENOTTY;
  		}
drivers/vfio/pci/vfio_pci_zdev.c +97 -1
index 4339f48b98bc..891cfa016d63 100644
@@ -13,6 +13,7 @@
  #include <linux/vfio_zdev.h>
  #include <asm/pci_clp.h>
  #include <asm/pci_io.h>
+#include <asm/pci_insn.h>
  #include <asm/kvm_pci.h>
 
  #include <linux/vfio_pci_core.h>
@@ -208,6 +209,99 @@ int vfio_pci_zdev_feat_interp(struct vfio_pci_core_device *vdev,
  	return rc;
  }
 
+int vfio_pci_zdev_feat_aif(struct vfio_pci_core_device *vdev,
+			   struct vfio_device_feature feature,
+			   unsigned long arg)
+{
+	struct zpci_dev *zdev = to_zpci(vdev->pdev);
+	struct vfio_device_zpci_aif *data;
+	struct vfio_device_feature *feat;
+	unsigned long minsz;
+	int size, rc = 0;
+
+	if (!zdev || !zdev->kzdev)
+		return -EINVAL;
+
+	/* If PROBE specified, return probe results immediately */
+	if (feature.flags & VFIO_DEVICE_FEATURE_PROBE)
+		return kvm_s390_pci_aif_probe(zdev);
+
+	/* GET and SET are mutually exclusive */
+	if ((feature.flags & VFIO_DEVICE_FEATURE_GET) &&
+	    (feature.flags & VFIO_DEVICE_FEATURE_SET))
+		return -EINVAL;
+
+	size = sizeof(*feat) + sizeof(*data);
+	feat = kzalloc(size, GFP_KERNEL);
+	if (!feat)
+		return -ENOMEM;
+
+	data = (struct vfio_device_zpci_aif *)&feat->data;
+	minsz = offsetofend(struct vfio_device_feature, flags);
+
+	if (feature.argsz < minsz + sizeof(*data))
+		return -EINVAL;
+
+	/* Get the rest of the payload for GET/SET */
+	rc = copy_from_user(data, (void __user *)(arg + minsz),
+			    sizeof(*data));
+	if (rc)
+		rc = -EINVAL;
+
+	if (feature.flags & VFIO_DEVICE_FEATURE_GET) {
+		if (zdev->kzdev->aif)
+			data->flags = VFIO_DEVICE_ZPCI_FLAG_AIF_FLOAT;
+		if (zdev->kzdev->fhost)
+			data->flags |= VFIO_DEVICE_ZPCI_FLAG_AIF_HOST;
+
+		if (copy_to_user((void __user *)arg, feat, size))
+			rc = -EFAULT;
+	} else if (feature.flags & VFIO_DEVICE_FEATURE_SET) {
+		if (data->flags & VFIO_DEVICE_ZPCI_FLAG_AIF_FLOAT) {
+			/* create a guest fib */
+			struct zpci_fib fib;
+
+			fib.fmt0.aibv = data->ibv;
+			fib.fmt0.isc = data->isc;
+			fib.fmt0.noi = data->noi;
+			if (data->sb != 0) {
+				fib.fmt0.aisb = data->sb;
+				fib.fmt0.aisbo = data->sbo;
+				fib.fmt0.sum = 1;
+			} else {
+				fib.fmt0.aisb = 0;
+				fib.fmt0.aisbo = 0;
+				fib.fmt0.sum = 0;
+			}
+			if (data->flags & VFIO_DEVICE_ZPCI_FLAG_AIF_HOST) {
+				rc = kvm_s390_pci_aif_enable(zdev, &fib, false);
+				if (!rc) {
+					zdev->kzdev->aif = true;
+					zdev->kzdev->fhost = true;
+				}
+			} else {
+				rc = kvm_s390_pci_aif_enable(zdev, &fib, true);
+				if (!rc)
+					zdev->kzdev->aif = true;
+			}
+		} else if (data->flags == 0) {
+			rc = kvm_s390_pci_aif_disable(zdev);
+			if (!rc) {
+				zdev->kzdev->aif = false;
+				zdev->kzdev->fhost = false;
+			}
+		} else {
+			rc = -EINVAL;
+		}
+	} else {
+		/* Neither GET nor SET were specified */
+		rc = -EINVAL;
+	}
+
+	kfree(feat);
+	return rc;
+}
+
  static int vfio_pci_zdev_group_notifier(struct notifier_block *nb,
  					unsigned long action, void *data)
  {
@@ -255,8 +349,10 @@ void vfio_pci_zdev_release(struct vfio_pci_core_device *vdev)
  	 * If the device was using interpretation, don't trust that userspace
  	 * did the appropriate cleanup
  	 */
-	if (zdev->gd != 0)
+	if (zdev->gd != 0) {
+		kvm_s390_pci_aif_disable(zdev);
  		kvm_s390_pci_interp_disable(zdev);
+	}
 
  	kvm_s390_pci_dev_release(zdev);
  }
include/linux/vfio_pci_core.h +10 -0
index 0db2b1051931..7ec5e82e7933 100644
@@ -201,6 +201,9 @@ extern int vfio_pci_info_zdev_add_caps(struct vfio_pci_core_device *vdev,
  int vfio_pci_zdev_feat_interp(struct vfio_pci_core_device *vdev,
  			      struct vfio_device_feature feature,
  			      unsigned long arg);
+int vfio_pci_zdev_feat_aif(struct vfio_pci_core_device *vdev,
+			   struct vfio_device_feature feature,
+			   unsigned long arg);
  void vfio_pci_zdev_open(struct vfio_pci_core_device *vdev);
  void vfio_pci_zdev_release(struct vfio_pci_core_device *vdev);
  #else
@@ -217,6 +220,13 @@ static inline int vfio_pci_zdev_feat_interp(struct vfio_pci_core_device *vdev,
  	return -ENOTTY;
  }
 
+static inline int vfio_pci_zdev_feat_aif(struct vfio_pci_core_device *vdev,
+					 struct vfio_device_feature feature,
+					 unsigned long arg)
+{
+	return -ENOTTY;
+}
+
  static inline void vfio_pci_zdev_open(struct vfio_pci_core_device *vdev)
  {
  }
include/uapi/linux/vfio.h +7 -0
index b9a75485b8e7..fe3bfd99bf50 100644
@@ -1009,6 +1009,13 @@ struct vfio_device_feature {
   */
  #define VFIO_DEVICE_FEATURE_ZPCI_INTERP		(1)
 
+/*
+ * Provide support for enbaling adapter interruption forwarding for zPCI
+ * devices.  This feature is only valid for s390x PCI devices.  Data provided
+ * when setting and getting this feature is further described in vfio_zdev.h
+ */
+#define VFIO_DEVICE_FEATURE_ZPCI_AIF		(2)
+
  /* -------- API for Type1 VFIO IOMMU -------- */
 
  /**
include/uapi/linux/vfio_zdev.h +3 -0
index 575f0410dc66..c574e23f9385 100644
@@ -90,4 +90,24 @@ struct vfio_device_zpci_interp {
  	__u32 fh;		/* Host device function handle */
  };
 
+/**
+ * VFIO_DEVICE_FEATURE_ZPCI_AIF
+ *

include/uapi/linux/vfio_zdev.h struct vfio_device_zpci_interp { · line 105

Quoted 10 lines
include/uapi/linux/vfio_zdev.h +10 -0
+ * This feature is used for enabling forwarding of adapter interrupts directly
+ * from firmware to the guest.  When setting this feature, the flags indicate
+ * whether to enable/disable the feature and the structure defined below is
+ * used to setup the forwarding structures.  When getting this feature, only
+ * the flags are used to indicate the current state.
+ */
+struct vfio_device_zpci_aif {
+	__u64 flags;
+#define VFIO_DEVICE_ZPCI_FLAG_AIF_FLOAT 1
+#define VFIO_DEVICE_ZPCI_FLAG_AIF_HOST 2

Generaly it looks good to me but I miss some explanation on these flags.

Which makes me realize that a more complete documentation under Documentation/S390 for VFIO zPCI as we have for VFIO AP and VFIO CCW would be of great interest.

Quoted 9 lines
include/uapi/linux/vfio_zdev.h +7 -0
+	__u64 ibv;		/* Address of guest interrupt bit vector */
+	__u64 sb;		/* Address of guest summary bit */
+	__u32 noi;		/* Number of interrupts */
+	__u8 isc;		/* Guest interrupt subclass */
+	__u8 sbo;		/* Offset of guest summary bit vector */
+};
+
  #endif
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#84 lore raw

Re: [PATCH v2 26/30] vfio-pci/zdev: wire up zPCI adapter interrupt forwarding support

Quoted 3 lines · Pierre Morel

On 1/14/22 21:31, Matthew Rosato wrote:

...

Quoted 18 lines
Quoted 18 lines
include/uapi/linux/vfio_zdev.h +0 -0
index 575f0410dc66..c574e23f9385 100644
@@ -90,4 +90,24 @@ struct vfio_device_zpci_interp {

      __u32 fh;        /* Host device function handle */
  };
+/**
+ * VFIO_DEVICE_FEATURE_ZPCI_AIF
+ *
+ * This feature is used for enabling forwarding of adapter interrupts directly
+ * from firmware to the guest.  When setting this feature, the flags indicate
+ * whether to enable/disable the feature and the structure defined below is
+ * used to setup the forwarding structures.  When getting this feature, only

Quoted 8 lines

include/uapi/linux/vfio_zdev.h struct vfio_device_zpci_interp { · line 105

Quoted 6 lines

+ * the flags are used to indicate the current state.
+ */

include/uapi/linux/vfio_zdev.h +4 -0
+struct vfio_device_zpci_aif {
+    __u64 flags;
+#define VFIO_DEVICE_ZPCI_FLAG_AIF_FLOAT 1
+#define VFIO_DEVICE_ZPCI_FLAG_AIF_HOST 2

Generaly it looks good to me but I miss some explanation on these flags.

I can add a small line comment for each, like:

AIF_FLOAT 1 /* Floating interrupts enabled */
AIF_HOST 2 /* Host delivery forced */

But here's a bit more detail:

On SET:
AIF_FLOAT = 1 means enable the interrupt forwarding assist for floating interrupt delivery AIF_FLOAT = 0 means to disable it.
AIF_HOST = 1 means the assist will always deliver the interrupt to the host and let the host inject it AIF_HOST = 0 host only gets interrupts when firmware can't deliver

on GET, we just indicate the current settings from the most recent SET, meaning:
AIF_FLOAT = 1 interrupt forwarding assist is currently active AIF_FLOAT = 0 interrupt forwarding assist is not currently active AIF_HOST = 1 interrupt forwarding will always go through host AIF_HOST = 0 interrupt forwarding will only go through the host when necessary

Quoted 2 lines

Which makes me realize that a more complete documentation under Documentation/S390 for VFIO zPCI as we have for VFIO AP and VFIO CCW would be of great interest.

You're not wrong -- a similar comment came up for QEMU. I will add this to my todo list as a follow-on.

#85 lore raw

Re: [PATCH v2 26/30] vfio-pci/zdev: wire up zPCI adapter interrupt forwarding support

In reply to Matthew Rosato
Quoted 21 lines · Matthew Rosato

Introduce support for VFIO_DEVICE_FEATURE_ZPCI_AIF, which is a new VFIO_DEVICE_FEATURE ioctl. This interface is used to indicate that an s390x vfio-pci device wishes to enable/disable zPCI adapter interrupt forwarding, which allows underlying firmware to deliver interrupts directly to the associated kvm guest.

---
  arch/s390/include/asm/kvm_pci.h  |  2 +
  drivers/vfio/pci/vfio_pci_core.c |  2 +
  drivers/vfio/pci/vfio_pci_zdev.c | 98 +++++++++++++++++++++++++++++++-
  include/linux/vfio_pci_core.h    | 10 ++++
  include/uapi/linux/vfio.h        |  7 +++
  include/uapi/linux/vfio_zdev.h   | 20 +++++++
  6 files changed, 138 insertions(+), 1 deletion(-)
arch/s390/include/asm/kvm_pci.h +0 -0
index dc00c3f27a00..dbab349a4a75 100644

arch/s390/include/asm/kvm_pci.h struct kvm_zdev { · line 40

Quoted 5 lines
arch/s390/include/asm/kvm_pci.h +2 -0
@@ -36,6 +36,8 @@ struct kvm_zdev {
  	struct zpci_fib fib;
  	struct notifier_block nb;
  	bool interp;
+	bool aif;
+	bool fhost;

Can we please have a comment on these booleans?
Can we have explicit naming to be able to follow their usage more easily? May be aif_float and aif_host to match with the VFIO feature?

Quoted 197 lines
arch/s390/include/asm/kvm_pci.h +0 -0
  };
 
  int kvm_s390_pci_dev_open(struct zpci_dev *zdev);
drivers/vfio/pci/vfio_pci_core.c +2 -0
index 2b2d64a2190c..01658de660bd 100644
@@ -1174,6 +1174,8 @@ long vfio_pci_core_ioctl(struct vfio_device *core_vdev, unsigned int cmd,
  			return 0;
  		case VFIO_DEVICE_FEATURE_ZPCI_INTERP:
  			return vfio_pci_zdev_feat_interp(vdev, feature, arg);
+		case VFIO_DEVICE_FEATURE_ZPCI_AIF:
+			return vfio_pci_zdev_feat_aif(vdev, feature, arg);
  		default:
  			return -ENOTTY;
  		}
drivers/vfio/pci/vfio_pci_zdev.c +97 -1
index 4339f48b98bc..891cfa016d63 100644
@@ -13,6 +13,7 @@
  #include <linux/vfio_zdev.h>
  #include <asm/pci_clp.h>
  #include <asm/pci_io.h>
+#include <asm/pci_insn.h>
  #include <asm/kvm_pci.h>
 
  #include <linux/vfio_pci_core.h>
@@ -208,6 +209,99 @@ int vfio_pci_zdev_feat_interp(struct vfio_pci_core_device *vdev,
  	return rc;
  }
 
+int vfio_pci_zdev_feat_aif(struct vfio_pci_core_device *vdev,
+			   struct vfio_device_feature feature,
+			   unsigned long arg)
+{
+	struct zpci_dev *zdev = to_zpci(vdev->pdev);
+	struct vfio_device_zpci_aif *data;
+	struct vfio_device_feature *feat;
+	unsigned long minsz;
+	int size, rc = 0;
+
+	if (!zdev || !zdev->kzdev)
+		return -EINVAL;
+
+	/* If PROBE specified, return probe results immediately */
+	if (feature.flags & VFIO_DEVICE_FEATURE_PROBE)
+		return kvm_s390_pci_aif_probe(zdev);
+
+	/* GET and SET are mutually exclusive */
+	if ((feature.flags & VFIO_DEVICE_FEATURE_GET) &&
+	    (feature.flags & VFIO_DEVICE_FEATURE_SET))
+		return -EINVAL;
+
+	size = sizeof(*feat) + sizeof(*data);
+	feat = kzalloc(size, GFP_KERNEL);
+	if (!feat)
+		return -ENOMEM;
+
+	data = (struct vfio_device_zpci_aif *)&feat->data;
+	minsz = offsetofend(struct vfio_device_feature, flags);
+
+	if (feature.argsz < minsz + sizeof(*data))
+		return -EINVAL;
+
+	/* Get the rest of the payload for GET/SET */
+	rc = copy_from_user(data, (void __user *)(arg + minsz),
+			    sizeof(*data));
+	if (rc)
+		rc = -EINVAL;
+
+	if (feature.flags & VFIO_DEVICE_FEATURE_GET) {
+		if (zdev->kzdev->aif)
+			data->flags = VFIO_DEVICE_ZPCI_FLAG_AIF_FLOAT;
+		if (zdev->kzdev->fhost)
+			data->flags |= VFIO_DEVICE_ZPCI_FLAG_AIF_HOST;
+
+		if (copy_to_user((void __user *)arg, feat, size))
+			rc = -EFAULT;
+	} else if (feature.flags & VFIO_DEVICE_FEATURE_SET) {
+		if (data->flags & VFIO_DEVICE_ZPCI_FLAG_AIF_FLOAT) {
+			/* create a guest fib */
+			struct zpci_fib fib;
+
+			fib.fmt0.aibv = data->ibv;
+			fib.fmt0.isc = data->isc;
+			fib.fmt0.noi = data->noi;
+			if (data->sb != 0) {
+				fib.fmt0.aisb = data->sb;
+				fib.fmt0.aisbo = data->sbo;
+				fib.fmt0.sum = 1;
+			} else {
+				fib.fmt0.aisb = 0;
+				fib.fmt0.aisbo = 0;
+				fib.fmt0.sum = 0;
+			}
+			if (data->flags & VFIO_DEVICE_ZPCI_FLAG_AIF_HOST) {
+				rc = kvm_s390_pci_aif_enable(zdev, &fib, false);
+				if (!rc) {
+					zdev->kzdev->aif = true;
+					zdev->kzdev->fhost = true;
+				}
+			} else {
+				rc = kvm_s390_pci_aif_enable(zdev, &fib, true);
+				if (!rc)
+					zdev->kzdev->aif = true;
+			}
+		} else if (data->flags == 0) {
+			rc = kvm_s390_pci_aif_disable(zdev);
+			if (!rc) {
+				zdev->kzdev->aif = false;
+				zdev->kzdev->fhost = false;
+			}
+		} else {
+			rc = -EINVAL;
+		}
+	} else {
+		/* Neither GET nor SET were specified */
+		rc = -EINVAL;
+	}
+
+	kfree(feat);
+	return rc;
+}
+
  static int vfio_pci_zdev_group_notifier(struct notifier_block *nb,
  					unsigned long action, void *data)
  {
@@ -255,8 +349,10 @@ void vfio_pci_zdev_release(struct vfio_pci_core_device *vdev)
  	 * If the device was using interpretation, don't trust that userspace
  	 * did the appropriate cleanup
  	 */
-	if (zdev->gd != 0)
+	if (zdev->gd != 0) {
+		kvm_s390_pci_aif_disable(zdev);
  		kvm_s390_pci_interp_disable(zdev);
+	}
 
  	kvm_s390_pci_dev_release(zdev);
  }
include/linux/vfio_pci_core.h +10 -0
index 0db2b1051931..7ec5e82e7933 100644
@@ -201,6 +201,9 @@ extern int vfio_pci_info_zdev_add_caps(struct vfio_pci_core_device *vdev,
  int vfio_pci_zdev_feat_interp(struct vfio_pci_core_device *vdev,
  			      struct vfio_device_feature feature,
  			      unsigned long arg);
+int vfio_pci_zdev_feat_aif(struct vfio_pci_core_device *vdev,
+			   struct vfio_device_feature feature,
+			   unsigned long arg);
  void vfio_pci_zdev_open(struct vfio_pci_core_device *vdev);
  void vfio_pci_zdev_release(struct vfio_pci_core_device *vdev);
  #else
@@ -217,6 +220,13 @@ static inline int vfio_pci_zdev_feat_interp(struct vfio_pci_core_device *vdev,
  	return -ENOTTY;
  }
 
+static inline int vfio_pci_zdev_feat_aif(struct vfio_pci_core_device *vdev,
+					 struct vfio_device_feature feature,
+					 unsigned long arg)
+{
+	return -ENOTTY;
+}
+
  static inline void vfio_pci_zdev_open(struct vfio_pci_core_device *vdev)
  {
  }
include/uapi/linux/vfio.h +7 -0
index b9a75485b8e7..fe3bfd99bf50 100644
@@ -1009,6 +1009,13 @@ struct vfio_device_feature {
   */
  #define VFIO_DEVICE_FEATURE_ZPCI_INTERP		(1)
 
+/*
+ * Provide support for enbaling adapter interruption forwarding for zPCI
+ * devices.  This feature is only valid for s390x PCI devices.  Data provided
+ * when setting and getting this feature is further described in vfio_zdev.h
+ */
+#define VFIO_DEVICE_FEATURE_ZPCI_AIF		(2)
+
  /* -------- API for Type1 VFIO IOMMU -------- */
 
  /**
include/uapi/linux/vfio_zdev.h +3 -0
index 575f0410dc66..c574e23f9385 100644
@@ -90,4 +90,24 @@ struct vfio_device_zpci_interp {
  	__u32 fh;		/* Host device function handle */
  };
 
+/**
+ * VFIO_DEVICE_FEATURE_ZPCI_AIF
+ *

include/uapi/linux/vfio_zdev.h struct vfio_device_zpci_interp { · line 105

Quoted 10 lines
include/uapi/linux/vfio_zdev.h +10 -0
+ * This feature is used for enabling forwarding of adapter interrupts directly
+ * from firmware to the guest.  When setting this feature, the flags indicate
+ * whether to enable/disable the feature and the structure defined below is
+ * used to setup the forwarding structures.  When getting this feature, only
+ * the flags are used to indicate the current state.
+ */
+struct vfio_device_zpci_aif {
+	__u64 flags;
+#define VFIO_DEVICE_ZPCI_FLAG_AIF_FLOAT 1
+#define VFIO_DEVICE_ZPCI_FLAG_AIF_HOST 2

I think we need more information on these flags.
What does AIF_FLOAT and what does AIF_HOST ?

Quoted 9 lines
include/uapi/linux/vfio_zdev.h +7 -0
+	__u64 ibv;		/* Address of guest interrupt bit vector */
+	__u64 sb;		/* Address of guest summary bit */
+	__u32 noi;		/* Number of interrupts */
+	__u8 isc;		/* Guest interrupt subclass */
+	__u8 sbo;		/* Offset of guest summary bit vector */
+};
+
  #endif
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#86 lore raw

Re: [PATCH v2 26/30] vfio-pci/zdev: wire up zPCI adapter interrupt forwarding support

Quoted 19 lines · Pierre Morel
Quoted 16 lines · Matthew Rosato

Introduce support for VFIO_DEVICE_FEATURE_ZPCI_AIF, which is a new VFIO_DEVICE_FEATURE ioctl.  This interface is used to indicate that an s390x vfio-pci device wishes to enable/disable zPCI adapter interrupt forwarding, which allows underlying firmware to deliver interrupts directly to the associated kvm guest.

---
  arch/s390/include/asm/kvm_pci.h  |  2 + drivers/vfio/pci/vfio_pci_core.c |  2 + drivers/vfio/pci/vfio_pci_zdev.c | 98 +++++++++++++++++++++++++++++++- include/linux/vfio_pci_core.h    | 10 ++++ include/uapi/linux/vfio.h        |  7 +++ include/uapi/linux/vfio_zdev.h   | 20 +++++++

  6 files changed, 138 insertions(+), 1 deletion(-)
Quoted 13 lines

arch/s390/include/asm/kvm_pci.h struct kvm_zdev { · line 40

Quoted 10 lines
arch/s390/include/asm/kvm_pci.h +2 -0
index dc00c3f27a00..dbab349a4a75 100644
@@ -36,6 +36,8 @@ struct kvm_zdev {
       struct zpci_fib fib;
       struct notifier_block nb;
       bool interp;
+    bool aif;
+    bool fhost;

Can we please have a comment on these booleans? > Can we have explicit naming to be able to follow their usage more easily? May be aif_float and aif_host to match with the VFIO feature?

Sure, rename would be fine.

As for a comment, maybe something like

bool aif_float; /* Enabled for floating interrupt assist */ bool aif_host; /* Require host delivery */

...

Quoted 19 lines
Quoted 19 lines
include/uapi/linux/vfio_zdev.h +0 -0
index 575f0410dc66..c574e23f9385 100644
@@ -90,4 +90,24 @@ struct vfio_device_zpci_interp {

      __u32 fh;        /* Host device function handle */
  };
+/**
+ * VFIO_DEVICE_FEATURE_ZPCI_AIF
+ *
+ * This feature is used for enabling forwarding of adapter interrupts directly
+ * from firmware to the guest.  When setting this feature, the flags indicate
+ * whether to enable/disable the feature and the structure defined below is
+ * used to setup the forwarding structures.  When getting this feature, only
+ * the flags are used to indicate the current state.

Quoted 9 lines

include/uapi/linux/vfio_zdev.h struct vfio_device_zpci_interp { · line 105

Quoted 5 lines

+ */

include/uapi/linux/vfio_zdev.h +4 -0
+struct vfio_device_zpci_aif {
+    __u64 flags;
+#define VFIO_DEVICE_ZPCI_FLAG_AIF_FLOAT 1
+#define VFIO_DEVICE_ZPCI_FLAG_AIF_HOST 2

I think we need more information on these flags.
What does AIF_FLOAT and what does AIF_HOST ?

You actually asked for this already on Jan 19 :), here's a copy of that response inline here:

I can add a small line comment for each, like:

AIF_FLOAT 1 /* Floating interrupts enabled */
AIF_HOST 2 /* Host delivery forced */

But here's a bit more detail:

On SET:
AIF_FLOAT = 1 means enable the interrupt forwarding assist for floating interrupt delivery AIF_FLOAT = 0 means to disable it.
AIF_HOST = 1 means the assist will always deliver the interrupt to the host and let the host inject it AIF_HOST = 0 host only gets interrupts when firmware can't deliver

on GET, we just indicate the current settings from the most recent SET, meaning:
AIF_FLOAT = 1 interrupt forwarding assist is currently active AIF_FLOAT = 0 interrupt forwarding assist is not currently active AIF_HOST = 1 interrupt forwarding will always go through host AIF_HOST = 0 interrupt forwarding will only go through the host when necessary

My thought would be add the line comments in this patch and then the additional detail in a follow-on patch that adds vfio zPCI to Documentation/S390

#87 lore raw

Re: [PATCH v2 26/30] vfio-pci/zdev: wire up zPCI adapter interrupt forwarding support

Quoted 34 lines · Matthew Rosato
Quoted 19 lines · Pierre Morel
Quoted 16 lines · Matthew Rosato

Introduce support for VFIO_DEVICE_FEATURE_ZPCI_AIF, which is a new VFIO_DEVICE_FEATURE ioctl.  This interface is used to indicate that an s390x vfio-pci device wishes to enable/disable zPCI adapter interrupt forwarding, which allows underlying firmware to deliver interrupts directly to the associated kvm guest.

---
  arch/s390/include/asm/kvm_pci.h  |  2 + drivers/vfio/pci/vfio_pci_core.c |  2 + drivers/vfio/pci/vfio_pci_zdev.c | 98 +++++++++++++++++++++++++++++++- include/linux/vfio_pci_core.h    | 10 ++++ include/uapi/linux/vfio.h        |  7 +++ include/uapi/linux/vfio_zdev.h   | 20 +++++++

  6 files changed, 138 insertions(+), 1 deletion(-)
Quoted 13 lines

arch/s390/include/asm/kvm_pci.h struct kvm_zdev { · line 40

Quoted 10 lines
arch/s390/include/asm/kvm_pci.h +2 -0
index dc00c3f27a00..dbab349a4a75 100644
@@ -36,6 +36,8 @@ struct kvm_zdev {
       struct zpci_fib fib;
       struct notifier_block nb;
       bool interp;
+    bool aif;
+    bool fhost;

Can we please have a comment on these booleans? > Can we have explicit naming to be able to follow their usage more easily? May be aif_float and aif_host to match with the VFIO feature?

Quoted 6 lines

Sure, rename would be fine.

As for a comment, maybe something like

bool aif_float; /* Enabled for floating interrupt assist */ bool aif_host;  /* Require host delivery */

good for me.

Quoted 32 lines

...

Quoted 19 lines
Quoted 19 lines
include/uapi/linux/vfio_zdev.h +0 -0
index 575f0410dc66..c574e23f9385 100644
@@ -90,4 +90,24 @@ struct vfio_device_zpci_interp {

      __u32 fh;        /* Host device function handle */ };
+/**
+ * VFIO_DEVICE_FEATURE_ZPCI_AIF
+ *
+ * This feature is used for enabling forwarding of adapter interrupts directly
+ * from firmware to the guest.  When setting this feature, the flags indicate
+ * whether to enable/disable the feature and the structure defined below is
+ * used to setup the forwarding structures.  When getting this feature, only
+ * the flags are used to indicate the current state.

Quoted 9 lines

include/uapi/linux/vfio_zdev.h struct vfio_device_zpci_interp { · line 105

Quoted 5 lines

+ */

include/uapi/linux/vfio_zdev.h +4 -0
+struct vfio_device_zpci_aif {
+    __u64 flags;
+#define VFIO_DEVICE_ZPCI_FLAG_AIF_FLOAT 1
+#define VFIO_DEVICE_ZPCI_FLAG_AIF_HOST 2

I think we need more information on these flags.
What does AIF_FLOAT and what does AIF_HOST ?

Quoted 1 line

You actually asked for this already on Jan 19 :), here's a copy of that response inline here:

:) I forgot

Quoted 14 lines

I can add a small line comment for each, like:

 AIF_FLOAT 1 /* Floating interrupts enabled */
 AIF_HOST 2  /* Host delivery forced */

But here's a bit more detail:

On SET:
AIF_FLOAT = 1 means enable the interrupt forwarding assist for floating interrupt delivery AIF_FLOAT = 0 means to disable it.
AIF_HOST = 1 means the assist will always deliver the interrupt to the host and let the host inject it AIF_HOST = 0 host only gets interrupts when firmware can't deliver

Quoted 8 lines

on GET, we just indicate the current settings from the most recent SET, meaning:
AIF_FLOAT = 1 interrupt forwarding assist is currently active AIF_FLOAT = 0 interrupt forwarding assist is not currently active AIF_HOST = 1 interrupt forwarding will always go through host AIF_HOST = 0 interrupt forwarding will only go through the host when necessary

My thought would be add the line comments in this patch and then the additional detail in a follow-on patch that adds vfio zPCI to Documentation/S390

good for me.

Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#88 lore raw

PATCH v2 27/30 vfio-pci/zdev: wire up zPCI IOAT assist support

In reply to Matthew Rosato

Introduce support for VFIO_DEVICE_FEATURE_ZPCI_IOAT, which is a new VFIO_DEVICE_FEATURE ioctl. This interface is used to indicate that an s390x vfio-pci device wishes to enable/disable zPCI I/O Address Translation assistance, allowing the host to perform address translation and shadowing.

---
 arch/s390/include/asm/kvm_pci.h  |  1 +
 drivers/vfio/pci/vfio_pci_core.c |  2 +
 drivers/vfio/pci/vfio_pci_zdev.c | 63 ++++++++++++++++++++++++++++++++
 include/linux/vfio_pci_core.h    | 10 +++++
 include/uapi/linux/vfio.h        |  8 ++++
 include/uapi/linux/vfio_zdev.h   | 13 +++++++
 6 files changed, 97 insertions(+)
arch/s390/include/asm/kvm_pci.h +1 -0
index dbab349a4a75..7b6b6d771026 100644
@@ -32,6 +32,7 @@ struct kvm_zdev {
 	struct zpci_dev *zdev;
 	struct kvm *kvm;
 	u64 rpcit_count;
+	u64 iota;
 	struct kvm_zdev_ioat ioat;
 	struct zpci_fib fib;
 	struct notifier_block nb;
drivers/vfio/pci/vfio_pci_core.c +2 -0
index 01658de660bd..709d9ba22a60 100644
@@ -1176,6 +1176,8 @@ long vfio_pci_core_ioctl(struct vfio_device *core_vdev, unsigned int cmd,
 			return vfio_pci_zdev_feat_interp(vdev, feature, arg);
 		case VFIO_DEVICE_FEATURE_ZPCI_AIF:
 			return vfio_pci_zdev_feat_aif(vdev, feature, arg);
+		case VFIO_DEVICE_FEATURE_ZPCI_IOAT:
+			return vfio_pci_zdev_feat_ioat(vdev, feature, arg);
 		default:
 			return -ENOTTY;
 		}
drivers/vfio/pci/vfio_pci_zdev.c +63 -0

75 lines not on this page · raw message

include/linux/vfio_pci_core.h +10 -0
index 7ec5e82e7933..f17d761ae14e 100644
@@ -204,6 +204,9 @@ int vfio_pci_zdev_feat_interp(struct vfio_pci_core_device *vdev,
 int vfio_pci_zdev_feat_aif(struct vfio_pci_core_device *vdev,
 			   struct vfio_device_feature feature,
 			   unsigned long arg);
+int vfio_pci_zdev_feat_ioat(struct vfio_pci_core_device *vdev,
+			    struct vfio_device_feature feature,
+			    unsigned long arg);
 void vfio_pci_zdev_open(struct vfio_pci_core_device *vdev);
 void vfio_pci_zdev_release(struct vfio_pci_core_device *vdev);
 #else
@@ -227,6 +230,13 @@ static inline int vfio_pci_zdev_feat_aif(struct vfio_pci_core_device *vdev,
 	return -ENOTTY;
 }
 
+static inline int vfio_pci_zdev_feat_ioat(struct vfio_pci_core_device *vdev,
+					  struct vfio_device_feature feature,
+					  unsigned long arg)
+{
+	return -ENOTTY;
+}
+
 static inline void vfio_pci_zdev_open(struct vfio_pci_core_device *vdev)
 {
 }
include/uapi/linux/vfio.h +8 -0
index fe3bfd99bf50..32c687388f48 100644
@@ -1016,6 +1016,14 @@ struct vfio_device_feature {
  */
 #define VFIO_DEVICE_FEATURE_ZPCI_AIF		(2)
 
+/*
+ * Provide support for enabling guest I/O address translation assistance for
+ * zPCI devices.  This feature is only valid for s390x PCI devices.  Data
+ * provided when setting and getting this feature is further described in
+ * vfio_zdev.h
+ */
+#define VFIO_DEVICE_FEATURE_ZPCI_IOAT		(3)
+
 /* -------- API for Type1 VFIO IOMMU -------- */
 
 /**
include/uapi/linux/vfio_zdev.h +13 -0
index c574e23f9385..1a5229b7bb18 100644
@@ -110,4 +110,17 @@ struct vfio_device_zpci_aif {
 	__u8 sbo;		/* Offset of guest summary bit vector */
 };
 
+/**
+ * VFIO_DEVICE_FEATURE_ZPCI_IOAT
+ *
+ * This feature is used for enabling guest I/O translation assistance for
+ * passthrough zPCI devices using instruction interpretation.  When setting
+ * this feature, the iota specifies a KVM guest I/O translation anchor.  When
+ * getting this feature, the most recently set anchor (or 0) is returned in
+ * iota.
+ */
+struct vfio_device_zpci_ioat {
+	__u64 iota;
+};
+
 #endif
Signature · 1 line
2.27.0
Reviewed-by on v2 27/30
#89 lore raw

Re: [PATCH v2 27/30] vfio-pci/zdev: wire up zPCI IOAT assist support

Quoted 6 lines · Matthew Rosato

Introduce support for VFIO_DEVICE_FEATURE_ZPCI_IOAT, which is a new VFIO_DEVICE_FEATURE ioctl. This interface is used to indicate that an s390x vfio-pci device wishes to enable/disable zPCI I/O Address Translation assistance, allowing the host to perform address translation and shadowing.

Quoted 40 lines
---
  arch/s390/include/asm/kvm_pci.h  |  1 +
  drivers/vfio/pci/vfio_pci_core.c |  2 +
  drivers/vfio/pci/vfio_pci_zdev.c | 63 ++++++++++++++++++++++++++++++++
  include/linux/vfio_pci_core.h    | 10 +++++
  include/uapi/linux/vfio.h        |  8 ++++
  include/uapi/linux/vfio_zdev.h   | 13 +++++++
  6 files changed, 97 insertions(+)
arch/s390/include/asm/kvm_pci.h +1 -0
index dbab349a4a75..7b6b6d771026 100644
@@ -32,6 +32,7 @@ struct kvm_zdev {
  	struct zpci_dev *zdev;
  	struct kvm *kvm;
  	u64 rpcit_count;
+	u64 iota;
  	struct kvm_zdev_ioat ioat;
  	struct zpci_fib fib;
  	struct notifier_block nb;
drivers/vfio/pci/vfio_pci_core.c +2 -0
index 01658de660bd..709d9ba22a60 100644
@@ -1176,6 +1176,8 @@ long vfio_pci_core_ioctl(struct vfio_device *core_vdev, unsigned int cmd,
  			return vfio_pci_zdev_feat_interp(vdev, feature, arg);
  		case VFIO_DEVICE_FEATURE_ZPCI_AIF:
  			return vfio_pci_zdev_feat_aif(vdev, feature, arg);
+		case VFIO_DEVICE_FEATURE_ZPCI_IOAT:
+			return vfio_pci_zdev_feat_ioat(vdev, feature, arg);
  		default:
  			return -ENOTTY;
  		}
drivers/vfio/pci/vfio_pci_zdev.c +0 -0
index 891cfa016d63..2b169d688937 100644

drivers/vfio/pci/vfio_pci_zdev.c int vfio_pci_zdev_feat_aif(struct vfio_pci_core_device *vdev, · line 310

Quoted 9 lines
drivers/vfio/pci/vfio_pci_zdev.c +6 -0
@@ -302,6 +302,68 @@ int vfio_pci_zdev_feat_aif(struct vfio_pci_core_device *vdev,
  	return rc;
  }
 
+int vfio_pci_zdev_feat_ioat(struct vfio_pci_core_device *vdev,
+			    struct vfio_device_feature feature,
+			    unsigned long arg)
+{
+	struct zpci_dev *zdev = to_zpci(vdev->pdev);
+	struct vfio_device_zpci_ioat *data;

NIT: something more explicit than "data" like "ioat_feature" ?

Quoted 137 lines
drivers/vfio/pci/vfio_pci_zdev.c +57 -0
+	struct vfio_device_feature *feat;
+	unsigned long minsz;
+	int size, rc = 0;
+
+	if (!zdev || !zdev->kzdev)
+		return -EINVAL;
+
+	/* If PROBE specified, return probe results immediately */
+	if (feature.flags & VFIO_DEVICE_FEATURE_PROBE)
+		return kvm_s390_pci_ioat_probe(zdev);
+
+	/* GET and SET are mutually exclusive */
+	if ((feature.flags & VFIO_DEVICE_FEATURE_GET) &&
+	    (feature.flags & VFIO_DEVICE_FEATURE_SET))
+		return -EINVAL;
+
+	size = sizeof(*feat) + sizeof(*data);
+	feat = kzalloc(size, GFP_KERNEL);
+	if (!feat)
+		return -ENOMEM;
+
+	data = (struct vfio_device_zpci_ioat *)&feat->data;
+	minsz = offsetofend(struct vfio_device_feature, flags);
+
+	if (feature.argsz < minsz + sizeof(*data))
+		return -EINVAL;
+
+	/* Get the rest of the payload for GET/SET */
+	rc = copy_from_user(data, (void __user *)(arg + minsz),
+			    sizeof(*data));
+	if (rc)
+		rc = -EINVAL;
+
+	if (feature.flags & VFIO_DEVICE_FEATURE_GET) {
+		data->iota = (u64)zdev->kzdev->iota;
+		if (copy_to_user((void __user *)arg, feat, size))
+			rc = -EFAULT;
+	} else if (feature.flags & VFIO_DEVICE_FEATURE_SET) {
+		if (data->iota != 0) {
+			rc = kvm_s390_pci_ioat_enable(zdev, data->iota);
+			if (!rc)
+				zdev->kzdev->iota = data->iota;
+		} else if (zdev->kzdev->iota != 0) {
+			rc = kvm_s390_pci_ioat_disable(zdev);
+			if (!rc)
+				zdev->kzdev->iota = 0;
+		}
+	} else {
+		/* Neither GET nor SET were specified */
+		rc = -EINVAL;
+	}
+
+	kfree(feat);
+	return rc;
+}
+
  static int vfio_pci_zdev_group_notifier(struct notifier_block *nb,
  					unsigned long action, void *data)
  {
@@ -351,6 +413,7 @@ void vfio_pci_zdev_release(struct vfio_pci_core_device *vdev)
  	 */
  	if (zdev->gd != 0) {
  		kvm_s390_pci_aif_disable(zdev);
+		kvm_s390_pci_ioat_disable(zdev);
  		kvm_s390_pci_interp_disable(zdev);
  	}
 
include/linux/vfio_pci_core.h +10 -0
index 7ec5e82e7933..f17d761ae14e 100644
@@ -204,6 +204,9 @@ int vfio_pci_zdev_feat_interp(struct vfio_pci_core_device *vdev,
  int vfio_pci_zdev_feat_aif(struct vfio_pci_core_device *vdev,
  			   struct vfio_device_feature feature,
  			   unsigned long arg);
+int vfio_pci_zdev_feat_ioat(struct vfio_pci_core_device *vdev,
+			    struct vfio_device_feature feature,
+			    unsigned long arg);
  void vfio_pci_zdev_open(struct vfio_pci_core_device *vdev);
  void vfio_pci_zdev_release(struct vfio_pci_core_device *vdev);
  #else
@@ -227,6 +230,13 @@ static inline int vfio_pci_zdev_feat_aif(struct vfio_pci_core_device *vdev,
  	return -ENOTTY;
  }
 
+static inline int vfio_pci_zdev_feat_ioat(struct vfio_pci_core_device *vdev,
+					  struct vfio_device_feature feature,
+					  unsigned long arg)
+{
+	return -ENOTTY;
+}
+
  static inline void vfio_pci_zdev_open(struct vfio_pci_core_device *vdev)
  {
  }
include/uapi/linux/vfio.h +8 -0
index fe3bfd99bf50..32c687388f48 100644
@@ -1016,6 +1016,14 @@ struct vfio_device_feature {
   */
  #define VFIO_DEVICE_FEATURE_ZPCI_AIF		(2)
 
+/*
+ * Provide support for enabling guest I/O address translation assistance for
+ * zPCI devices.  This feature is only valid for s390x PCI devices.  Data
+ * provided when setting and getting this feature is further described in
+ * vfio_zdev.h
+ */
+#define VFIO_DEVICE_FEATURE_ZPCI_IOAT		(3)
+
  /* -------- API for Type1 VFIO IOMMU -------- */
 
  /**
include/uapi/linux/vfio_zdev.h +13 -0
index c574e23f9385..1a5229b7bb18 100644
@@ -110,4 +110,17 @@ struct vfio_device_zpci_aif {
  	__u8 sbo;		/* Offset of guest summary bit vector */
  };
 
+/**
+ * VFIO_DEVICE_FEATURE_ZPCI_IOAT
+ *
+ * This feature is used for enabling guest I/O translation assistance for
+ * passthrough zPCI devices using instruction interpretation.  When setting
+ * this feature, the iota specifies a KVM guest I/O translation anchor.  When
+ * getting this feature, the most recently set anchor (or 0) is returned in
+ * iota.
+ */
+struct vfio_device_zpci_ioat {
+	__u64 iota;
+};
+
  #endif
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#90 lore raw

PATCH v2 28/30 vfio-pci/zdev: add DTSM to clp group capability

In reply to Matthew Rosato

The DTSM, or designation type supported mask, indicates what IOAT formats are available to the guest. For an interpreted device, userspace will not know what format(s) the IOAT assist supports, so pass it via the capability chain. Since the value belongs to the Query PCI Function Group clp, let's extend the existing capability with a new version.

---
 drivers/vfio/pci/vfio_pci_zdev.c | 9 ++++++---
 include/uapi/linux/vfio_zdev.h   | 3 +++
 2 files changed, 9 insertions(+), 3 deletions(-)
drivers/vfio/pci/vfio_pci_zdev.c +6 -3
index 2b169d688937..aa2ef9067c7d 100644
@@ -45,19 +45,22 @@ static int zpci_group_cap(struct zpci_dev *zdev, struct vfio_info_cap *caps)
 {
 	struct vfio_device_info_cap_zpci_group cap = {
 		.header.id = VFIO_DEVICE_INFO_CAP_ZPCI_GROUP,
-		.header.version = 1,
+		.header.version = 2,
 		.dasm = zdev->dma_mask,
 		.msi_addr = zdev->msi_addr,
 		.flags = VFIO_DEVICE_INFO_ZPCI_FLAG_REFRESH,
 		.mui = zdev->fmb_update,
 		.noi = zdev->max_msi,
 		.maxstbl = ZPCI_MAX_WRITE_SIZE,
-		.version = zdev->version
+		.version = zdev->version,
+		.dtsm = 0
 	};
 
 	/* Some values are different for interpreted devices */
-	if (zdev->kzdev && zdev->kzdev->interp)
+	if (zdev->kzdev && zdev->kzdev->interp) {
 		cap.maxstbl = zdev->maxstbl;
+		cap.dtsm = kvm_s390_pci_get_dtsm(zdev);
+	}
 
 	return vfio_info_add_capability(caps, &cap.header, sizeof(cap));
 }
include/uapi/linux/vfio_zdev.h +3 -0
index 1a5229b7bb18..b4c2ba8e71f0 100644
@@ -47,6 +47,9 @@ struct vfio_device_info_cap_zpci_group {
 	__u16 noi;		/* Maximum number of MSIs */
 	__u16 maxstbl;		/* Maximum Store Block Length */
 	__u8 version;		/* Supported PCI Version */
+	/* End of version 1 */
+	__u8 dtsm;		/* Supported IOAT Designations */
+	/* End of version 2 */
 };
 
 /**
Signature · 1 line
2.27.0
Reviewed-by on v2 28/30
#91 lore raw

Re: [PATCH v2 28/30] vfio-pci/zdev: add DTSM to clp group capability

Quoted 7 lines · Matthew Rosato

The DTSM, or designation type supported mask, indicates what IOAT formats are available to the guest. For an interpreted device, userspace will not know what format(s) the IOAT assist supports, so pass it via the capability chain. Since the value belongs to the Query PCI Function Group clp, let's extend the existing capability with a new version.

Quoted 50 lines
---
  drivers/vfio/pci/vfio_pci_zdev.c | 9 ++++++---
  include/uapi/linux/vfio_zdev.h   | 3 +++
  2 files changed, 9 insertions(+), 3 deletions(-)
drivers/vfio/pci/vfio_pci_zdev.c +6 -3
index 2b169d688937..aa2ef9067c7d 100644
@@ -45,19 +45,22 @@ static int zpci_group_cap(struct zpci_dev *zdev, struct vfio_info_cap *caps)
  {
  	struct vfio_device_info_cap_zpci_group cap = {
  		.header.id = VFIO_DEVICE_INFO_CAP_ZPCI_GROUP,
-		.header.version = 1,
+		.header.version = 2,
  		.dasm = zdev->dma_mask,
  		.msi_addr = zdev->msi_addr,
  		.flags = VFIO_DEVICE_INFO_ZPCI_FLAG_REFRESH,
  		.mui = zdev->fmb_update,
  		.noi = zdev->max_msi,
  		.maxstbl = ZPCI_MAX_WRITE_SIZE,
-		.version = zdev->version
+		.version = zdev->version,
+		.dtsm = 0
  	};
 
  	/* Some values are different for interpreted devices */
-	if (zdev->kzdev && zdev->kzdev->interp)
+	if (zdev->kzdev && zdev->kzdev->interp) {
  		cap.maxstbl = zdev->maxstbl;
+		cap.dtsm = kvm_s390_pci_get_dtsm(zdev);
+	}
 
  	return vfio_info_add_capability(caps, &cap.header, sizeof(cap));
  }
include/uapi/linux/vfio_zdev.h +3 -0
index 1a5229b7bb18..b4c2ba8e71f0 100644
@@ -47,6 +47,9 @@ struct vfio_device_info_cap_zpci_group {
  	__u16 noi;		/* Maximum number of MSIs */
  	__u16 maxstbl;		/* Maximum Store Block Length */
  	__u8 version;		/* Supported PCI Version */
+	/* End of version 1 */
+	__u8 dtsm;		/* Supported IOAT Designations */
+	/* End of version 2 */
  };
 
  /**
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#92 lore raw

PATCH v2 29/30 KVM: s390: introduce CPU feature for zPCI Interpretation

In reply to Matthew Rosato

KVM_S390_VM_CPU_FEAT_ZPCI_INTERP relays whether zPCI interpretive execution is possible based on the available hardware facilities.

---
 arch/s390/include/uapi/asm/kvm.h | 1 +
 arch/s390/kvm/kvm-s390.c         | 4 ++++
 2 files changed, 5 insertions(+)
arch/s390/include/uapi/asm/kvm.h +1 -0
index 7a6b14874d65..ed06458a871f 100644
@@ -130,6 +130,7 @@ struct kvm_s390_vm_cpu_machine {
 #define KVM_S390_VM_CPU_FEAT_PFMFI	11
 #define KVM_S390_VM_CPU_FEAT_SIGPIF	12
 #define KVM_S390_VM_CPU_FEAT_KSS	13
+#define KVM_S390_VM_CPU_FEAT_ZPCI_INTERP 14
 struct kvm_s390_vm_cpu_feat {
 	__u64 feat[16];
 };
arch/s390/kvm/kvm-s390.c +4 -0
index b6c32fc3b272..3ed59fe512dd 100644
@@ -434,6 +434,10 @@ static void kvm_s390_cpu_feat_init(void)
 	if (test_facility(151)) /* DFLTCC */
 		__insn32_query(INSN_DFLTCC, kvm_s390_available_subfunc.dfltcc);
 
+	if (test_facility(69) && test_facility(70) && test_facility(71) &&
+	    test_facility(72)) /* zPCI Interpretation */
+		allow_cpu_feat(KVM_S390_VM_CPU_FEAT_ZPCI_INTERP);
+
 	if (MACHINE_HAS_ESOP)
 		allow_cpu_feat(KVM_S390_VM_CPU_FEAT_ESOP);
 	/*
Signature · 1 line
2.27.0
#93 lore raw

Re: [PATCH v2 29/30] KVM: s390: introduce CPU feature for zPCI Interpretation

Quoted 26 lines · Matthew Rosato

KVM_S390_VM_CPU_FEAT_ZPCI_INTERP relays whether zPCI interpretive execution is possible based on the available hardware facilities.

---
  arch/s390/include/uapi/asm/kvm.h | 1 +
  arch/s390/kvm/kvm-s390.c         | 4 ++++
  2 files changed, 5 insertions(+)
arch/s390/include/uapi/asm/kvm.h +1 -0
index 7a6b14874d65..ed06458a871f 100644
@@ -130,6 +130,7 @@ struct kvm_s390_vm_cpu_machine {
  #define KVM_S390_VM_CPU_FEAT_PFMFI	11
  #define KVM_S390_VM_CPU_FEAT_SIGPIF	12
  #define KVM_S390_VM_CPU_FEAT_KSS	13
+#define KVM_S390_VM_CPU_FEAT_ZPCI_INTERP 14
  struct kvm_s390_vm_cpu_feat {
  	__u64 feat[16];
  };
arch/s390/kvm/kvm-s390.c +0 -0
index b6c32fc3b272..3ed59fe512dd 100644

arch/s390/kvm/kvm-s390.c static void kvm_s390_cpu_feat_init(void) · line 440

Quoted 7 lines
arch/s390/kvm/kvm-s390.c +4 -0
@@ -434,6 +434,10 @@ static void kvm_s390_cpu_feat_init(void)
  	if (test_facility(151)) /* DFLTCC */
  		__insn32_query(INSN_DFLTCC, kvm_s390_available_subfunc.dfltcc);
 
+	if (test_facility(69) && test_facility(70) && test_facility(71) &&
+	    test_facility(72)) /* zPCI Interpretation */
+		allow_cpu_feat(KVM_S390_VM_CPU_FEAT_ZPCI_INTERP);
+

Don't we want to start the support of ZPCI interpretation starting with Z14 ?

Quoted 4 lines
arch/s390/kvm/kvm-s390.c +0 -0
  	if (MACHINE_HAS_ESOP)
  		allow_cpu_feat(KVM_S390_VM_CPU_FEAT_ESOP);
  	/*
Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen
author maintainer
#94 lore raw

PATCH v2 30/30 MAINTAINERS: additional files related kvm s390 pci passthrough

In reply to Matthew Rosato

Add entries from the s390 kvm subdirectory related to pci passthrough.

---
 MAINTAINERS | 2 ++
 1 file changed, 2 insertions(+)
MAINTAINERS +2 -0
index 5d0cd537803a..1b52acd74cfd 100644
@@ -16874,6 +16874,8 @@ M:	Eric Farman <[email protected]>
 L:	[email protected]
 L:	[email protected]
 S:	Supported
+F:	arch/s390/include/asm/kvm_pci.h
+F:	arch/s390/kvm/pci*
 F:	drivers/vfio/pci/vfio_pci_zdev.c
 F:	include/uapi/linux/vfio_zdev.h
 
Signature · 1 line
2.27.0
author maintainer
#95 lore raw
In reply to Matthew Rosato
Quoted 10 lines · Matthew Rosato

Enable interpretive execution of zPCI instructions + adapter interruption forwarding for s390x KVM vfio-pci. This is done by introducing a series of new vfio-pci feature ioctls that are unique vfio-pci-zdev (s390x) and are used to negotiate the various aspects of zPCI interpretation setup. By allowing intepretation of zPCI instructions and firmware delivery of interrupts to guests, we can significantly reduce the frequency of guest SIE exits for zPCI. We then see additional gains by handling a hot-path instruction that can still intercept to the hypervisor (RPCIT) directly in kvm.

Quoted 5 lines

From the perspective of guest configuration, you passthrough zPCI devices
in the same manner as before, with intepretation support being used by default if available in kernel+qemu.

Will reply with a link to the associated QEMU series.

https://lore.kernel.org/qemu-devel/[email protected]/

#96 lore raw
In reply to Matthew Rosato
Quoted 10 lines · Matthew Rosato

Enable interpretive execution of zPCI instructions + adapter interruption forwarding for s390x KVM vfio-pci. This is done by introducing a series of new vfio-pci feature ioctls that are unique vfio-pci-zdev (s390x) and are used to negotiate the various aspects of zPCI interpretation setup. By allowing intepretation of zPCI instructions and firmware delivery of interrupts to guests, we can significantly reduce the frequency of guest SIE exits for zPCI. We then see additional gains by handling a hot-path instruction that can still intercept to the hypervisor (RPCIT) directly in kvm.

Quoted 5 lines

From the perspective of guest configuration, you passthrough zPCI devices
in the same manner as before, with intepretation support being used by default if available in kernel+qemu.

Will reply with a link to the associated QEMU series.

I did the comment in a patch but I think that centralizing it here is clearer: I think having a documentation in Documentation/S390 like we have already for VFIO AP and VFIO CCW would be a good thing.

Signature · 2 lines
Pierre Morel
IBM Lab Boeblingen