All of lore.kernel.org
 help / color / mirror / Atom feed
* [Qemu-devel] [PATCH v2 0/3] s390x/sclp: 64 bit event masks
@ 2018-02-22 16:22 Claudio Imbrenda
  2018-02-22 16:22 ` [Qemu-devel] [PATCH v2 1/3] s390x/sclp: proper support of larger send and receive masks Claudio Imbrenda
                   ` (3 more replies)
  0 siblings, 4 replies; 8+ messages in thread
From: Claudio Imbrenda @ 2018-02-22 16:22 UTC (permalink / raw)
  To: cohuck; +Cc: qemu-s390x, qemu-devel, borntraeger

Until 67915de9f0383ccf4a ("s390x/event-facility: variable-length event masks")
we only supported 32bit sclp event masks, even though the archiecture
allows the guests to set up sclp event masks up to 1021 bytes in length.
With that patch the behaviour was almost compliant, but some issues were
still remaining, in particular regarding the handling of selective reads
and migration.

This patchset fixes migration and the handling of selective reads, and
puts in place the support for 64-bit sclp event masks internally.

A new property of the sclp-event device switches between the 32bit masks
and the compliant behaviour. The property is bound to the machine
version, so older machines keep the old broken behaviour, allowing for
migration, but the default is the compliant implementation.

Fixes: 67915de9f0383ccf4a ("s390x/event-facility: variable-length event masks")

v1 -> v2 
* improved comments and patch descriptions to better explain why we need
  this, including better description of the old broken behaviour
* rename SCLPEVMSK to SCLP_EVMASK to improve readability
* removed some unneded variable initializations
* fixed a pre-existing typo

Claudio Imbrenda (3):
  s390x/sclp: proper support of larger send and receive masks
  s390x/sclp: clean up sclp masks
  s390x/sclp: extend SCLP event masks to 64 bits

 hw/char/sclpconsole-lm.c          |   4 +-
 hw/char/sclpconsole.c             |   4 +-
 hw/s390x/event-facility.c         | 150 +++++++++++++++++++++++++++++++-------
 hw/s390x/s390-virtio-ccw.c        |   8 +-
 hw/s390x/sclpcpu.c                |   4 +-
 hw/s390x/sclpquiesce.c            |   4 +-
 include/hw/s390x/event-facility.h |  22 +++---
 7 files changed, 151 insertions(+), 45 deletions(-)

-- 
2.7.4

^ permalink raw reply	[flat|nested] 8+ messages in thread

* [Qemu-devel] [PATCH v2 1/3] s390x/sclp: proper support of larger send and receive masks
  2018-02-22 16:22 [Qemu-devel] [PATCH v2 0/3] s390x/sclp: 64 bit event masks Claudio Imbrenda
@ 2018-02-22 16:22 ` Claudio Imbrenda
  2018-02-23 10:31   ` Cornelia Huck
  2018-02-22 16:22 ` [Qemu-devel] [PATCH v2 2/3] s390x/sclp: clean up sclp masks Claudio Imbrenda
                   ` (2 subsequent siblings)
  3 siblings, 1 reply; 8+ messages in thread
From: Claudio Imbrenda @ 2018-02-22 16:22 UTC (permalink / raw)
  To: cohuck; +Cc: qemu-s390x, qemu-devel, borntraeger

Until 67915de9f0383ccf4a ("s390x/event-facility: variable-length event masks")
we only supported sclp event masks of size exactly 4 bytes, even though
the archiecture allows the guests to set up sclp event masks from 1 to
1021 bytes in length.
After that patch, the behaviour was almost compliant, but some issues
were still remaining, in particular regarding the handling of selective
reads and migration.

When setting the sclp event mask, a mask size is also specified. Until
now we only considered the size in order to decide which bits to save
in the internal state. On the other hand, when a guest performs a
selective read, it sends a mask, but it does not specify a size; the
implied size is the size of the last mask that has been set.

Specifying bits in the mask of selective read that are not available in
the internal mask should return an error, and bits past the end of the
mask should obviously be ignored. This can only be achieved by keeping
track of the lenght of the mask.

The mask length is thus now part of the internal state that needs to be
migrated.

This patch fixes the handling of selective reads, whose size will now
match the length of the event mask, as per architecture.

While the default behaviour is to be compliant with the architecture,
when using older machine models the old broken behaviour is selected
(allowing only masks of size exactly 4), in order to be able to migrate
toward older versions.

Fixes: 67915de9f0383ccf4a ("s390x/event-facility: variable-length event masks")
Signed-off-by: Claudio Imbrenda <imbrenda@linux.vnet.ibm.com>
---
 hw/s390x/event-facility.c  | 91 +++++++++++++++++++++++++++++++++++++++-------
 hw/s390x/s390-virtio-ccw.c |  8 +++-
 2 files changed, 85 insertions(+), 14 deletions(-)

diff --git a/hw/s390x/event-facility.c b/hw/s390x/event-facility.c
index 155a694..b7cd075 100644
--- a/hw/s390x/event-facility.c
+++ b/hw/s390x/event-facility.c
@@ -31,6 +31,15 @@ struct SCLPEventFacility {
     SCLPEventsBus sbus;
     /* guest' receive mask */
     unsigned int receive_mask;
+    /*
+     * when false, we keep the same broken, backwards compatible behaviour as
+     * before, allowing only masks of size exactly 4; when true, we implement
+     * the architecture correctly, allowing all valid mask sizes. Needed for
+     * migration toward older versions.
+     */
+    bool allow_all_mask_sizes;
+    /* length of the receive mask */
+    uint16_t mask_length;
 };
 
 /* return true if any child has event pending set */
@@ -220,6 +229,17 @@ static uint16_t handle_sccb_read_events(SCLPEventFacility *ef, SCCB *sccb,
     return rc;
 }
 
+/* copy up to dst_len bytes and fill the rest of dst with zeroes */
+static void copy_mask(uint8_t *dst, uint8_t *src, uint16_t dst_len,
+                      uint16_t src_len)
+{
+    int i;
+
+    for (i = 0; i < dst_len; i++) {
+        dst[i] = i < src_len ? src[i] : 0;
+    }
+}
+
 static void read_event_data(SCLPEventFacility *ef, SCCB *sccb)
 {
     unsigned int sclp_active_selection_mask;
@@ -240,7 +260,9 @@ static void read_event_data(SCLPEventFacility *ef, SCCB *sccb)
         sclp_active_selection_mask = sclp_cp_receive_mask;
         break;
     case SCLP_SELECTIVE_READ:
-        sclp_active_selection_mask = be32_to_cpu(red->mask);
+        copy_mask((uint8_t *)&sclp_active_selection_mask, (uint8_t *)&red->mask,
+                  sizeof(sclp_active_selection_mask), ef->mask_length);
+        sclp_active_selection_mask = be32_to_cpu(sclp_active_selection_mask);
         if (!sclp_cp_receive_mask ||
             (sclp_active_selection_mask & ~sclp_cp_receive_mask)) {
             sccb->h.response_code =
@@ -259,24 +281,14 @@ out:
     return;
 }
 
-/* copy up to dst_len bytes and fill the rest of dst with zeroes */
-static void copy_mask(uint8_t *dst, uint8_t *src, uint16_t dst_len,
-                      uint16_t src_len)
-{
-    int i;
-
-    for (i = 0; i < dst_len; i++) {
-        dst[i] = i < src_len ? src[i] : 0;
-    }
-}
-
 static void write_event_mask(SCLPEventFacility *ef, SCCB *sccb)
 {
     WriteEventMask *we_mask = (WriteEventMask *) sccb;
     uint16_t mask_length = be16_to_cpu(we_mask->mask_length);
     uint32_t tmp_mask;
 
-    if (!mask_length || (mask_length > SCLP_EVENT_MASK_LEN_MAX)) {
+    if (!mask_length || (mask_length > SCLP_EVENT_MASK_LEN_MAX) ||
+        ((mask_length != 4) && !ef->allow_all_mask_sizes)) {
         sccb->h.response_code = cpu_to_be16(SCLP_RC_INVALID_MASK_LENGTH);
         goto out;
     }
@@ -301,6 +313,7 @@ static void write_event_mask(SCLPEventFacility *ef, SCCB *sccb)
               mask_length, sizeof(tmp_mask));
 
     sccb->h.response_code = cpu_to_be16(SCLP_RC_NORMAL_COMPLETION);
+    ef->mask_length = mask_length;
 
 out:
     return;
@@ -356,6 +369,34 @@ static void command_handler(SCLPEventFacility *ef, SCCB *sccb, uint64_t code)
     }
 }
 
+static bool vmstate_event_facility_mask_length_needed(void *opaque)
+{
+    SCLPEventFacility *ef = opaque;
+
+    return ef->allow_all_mask_sizes;
+}
+
+static int vmstate_event_facility_mask_length_pre_load(void *opaque)
+{
+    SCLPEventFacility *ef = opaque;
+
+    ef->allow_all_mask_sizes = false;
+    return 0;
+}
+
+static const VMStateDescription vmstate_event_facility_mask_length = {
+    .name = "vmstate-event-facility/mask_length",
+    .version_id = 0,
+    .minimum_version_id = 0,
+    .needed = vmstate_event_facility_mask_length_needed,
+    .pre_load = vmstate_event_facility_mask_length_pre_load,
+    .fields = (VMStateField[]) {
+        VMSTATE_BOOL(allow_all_mask_sizes, SCLPEventFacility),
+        VMSTATE_UINT16(mask_length, SCLPEventFacility),
+        VMSTATE_END_OF_LIST()
+     }
+};
+
 static const VMStateDescription vmstate_event_facility = {
     .name = "vmstate-event-facility",
     .version_id = 0,
@@ -363,15 +404,39 @@ static const VMStateDescription vmstate_event_facility = {
     .fields = (VMStateField[]) {
         VMSTATE_UINT32(receive_mask, SCLPEventFacility),
         VMSTATE_END_OF_LIST()
+     },
+    .subsections = (const VMStateDescription * []) {
+        &vmstate_event_facility_mask_length,
+        NULL
      }
 };
 
+static void sclp_event_set_allow_all_mask_sizes(Object *obj, bool value,
+                                                       Error **errp)
+{
+    SCLPEventFacility *ef = (SCLPEventFacility *)obj;
+
+    ef->allow_all_mask_sizes = value;
+}
+
+static bool sclp_event_get_allow_all_mask_sizes(Object *obj, Error **e)
+{
+    SCLPEventFacility *ef = (SCLPEventFacility *)obj;
+
+    return ef->allow_all_mask_sizes;
+}
+
 static void init_event_facility(Object *obj)
 {
     SCLPEventFacility *event_facility = EVENT_FACILITY(obj);
     DeviceState *sdev = DEVICE(obj);
     Object *new;
 
+    event_facility->mask_length = 4;
+    event_facility->allow_all_mask_sizes = true;
+    object_property_add_bool(obj, "allow_all_mask_sizes",
+                             sclp_event_get_allow_all_mask_sizes,
+                             sclp_event_set_allow_all_mask_sizes, NULL);
     /* Spawn a new bus for SCLP events */
     qbus_create_inplace(&event_facility->sbus, sizeof(event_facility->sbus),
                         TYPE_SCLP_EVENTS_BUS, sdev, NULL);
diff --git a/hw/s390x/s390-virtio-ccw.c b/hw/s390x/s390-virtio-ccw.c
index 8b3053f..e9309fd 100644
--- a/hw/s390x/s390-virtio-ccw.c
+++ b/hw/s390x/s390-virtio-ccw.c
@@ -29,6 +29,7 @@
 #include "s390-pci-bus.h"
 #include "hw/s390x/storage-keys.h"
 #include "hw/s390x/storage-attributes.h"
+#include "hw/s390x/event-facility.h"
 #include "hw/compat.h"
 #include "ipl.h"
 #include "hw/s390x/s390-virtio-ccw.h"
@@ -671,7 +672,12 @@ bool css_migration_enabled(void)
     type_init(ccw_machine_register_##suffix)
 
 #define CCW_COMPAT_2_11 \
-        HW_COMPAT_2_11
+        HW_COMPAT_2_11 \
+        {\
+            .driver   = TYPE_SCLP_EVENT_FACILITY,\
+            .property = "allow_all_mask_sizes",\
+            .value    = "off",\
+        },
 
 #define CCW_COMPAT_2_10 \
         HW_COMPAT_2_10
-- 
2.7.4

^ permalink raw reply related	[flat|nested] 8+ messages in thread

* [Qemu-devel] [PATCH v2 2/3] s390x/sclp: clean up sclp masks
  2018-02-22 16:22 [Qemu-devel] [PATCH v2 0/3] s390x/sclp: 64 bit event masks Claudio Imbrenda
  2018-02-22 16:22 ` [Qemu-devel] [PATCH v2 1/3] s390x/sclp: proper support of larger send and receive masks Claudio Imbrenda
@ 2018-02-22 16:22 ` Claudio Imbrenda
  2018-02-22 16:22 ` [Qemu-devel] [PATCH v2 3/3] s390x/sclp: extend SCLP event masks to 64 bits Claudio Imbrenda
  2018-02-23 10:37 ` [Qemu-devel] [PATCH v2 0/3] s390x/sclp: 64 bit event masks Cornelia Huck
  3 siblings, 0 replies; 8+ messages in thread
From: Claudio Imbrenda @ 2018-02-22 16:22 UTC (permalink / raw)
  To: cohuck; +Cc: qemu-s390x, qemu-devel, borntraeger

Introduce an sccb_mask_t to be used for SCLP event masks instead of just
unsigned int or uint32_t. This will allow later to extend the mask with
more ease.

Signed-off-by: Claudio Imbrenda <imbrenda@linux.vnet.ibm.com>
---
 hw/char/sclpconsole-lm.c          |  4 ++--
 hw/char/sclpconsole.c             |  4 ++--
 hw/s390x/event-facility.c         | 20 ++++++++++----------
 hw/s390x/sclpcpu.c                |  4 ++--
 hw/s390x/sclpquiesce.c            |  4 ++--
 include/hw/s390x/event-facility.h | 22 +++++++++++++---------
 6 files changed, 31 insertions(+), 27 deletions(-)

diff --git a/hw/char/sclpconsole-lm.c b/hw/char/sclpconsole-lm.c
index c500bda..cc4d70a 100644
--- a/hw/char/sclpconsole-lm.c
+++ b/hw/char/sclpconsole-lm.c
@@ -102,12 +102,12 @@ static bool can_handle_event(uint8_t type)
     return type == SCLP_EVENT_MESSAGE || type == SCLP_EVENT_PMSGCMD;
 }
 
-static unsigned int send_mask(void)
+static sccb_mask_t send_mask(void)
 {
     return SCLP_EVENT_MASK_OP_CMD | SCLP_EVENT_MASK_PMSGCMD;
 }
 
-static unsigned int receive_mask(void)
+static sccb_mask_t receive_mask(void)
 {
     return SCLP_EVENT_MASK_MSG | SCLP_EVENT_MASK_PMSGCMD;
 }
diff --git a/hw/char/sclpconsole.c b/hw/char/sclpconsole.c
index d0265df..ec9db13 100644
--- a/hw/char/sclpconsole.c
+++ b/hw/char/sclpconsole.c
@@ -83,12 +83,12 @@ static bool can_handle_event(uint8_t type)
     return type == SCLP_EVENT_ASCII_CONSOLE_DATA;
 }
 
-static unsigned int send_mask(void)
+static sccb_mask_t send_mask(void)
 {
     return SCLP_EVENT_MASK_MSG_ASCII;
 }
 
-static unsigned int receive_mask(void)
+static sccb_mask_t receive_mask(void)
 {
     return SCLP_EVENT_MASK_MSG_ASCII;
 }
diff --git a/hw/s390x/event-facility.c b/hw/s390x/event-facility.c
index b7cd075..94fe948 100644
--- a/hw/s390x/event-facility.c
+++ b/hw/s390x/event-facility.c
@@ -29,8 +29,8 @@ typedef struct SCLPEventsBus {
 struct SCLPEventFacility {
     SysBusDevice parent_obj;
     SCLPEventsBus sbus;
-    /* guest' receive mask */
-    unsigned int receive_mask;
+    /* guest's receive mask */
+    sccb_mask_t receive_mask;
     /*
      * when false, we keep the same broken, backwards compatible behaviour as
      * before, allowing only masks of size exactly 4; when true, we implement
@@ -61,9 +61,9 @@ static bool event_pending(SCLPEventFacility *ef)
     return false;
 }
 
-static unsigned int get_host_send_mask(SCLPEventFacility *ef)
+static sccb_mask_t get_host_send_mask(SCLPEventFacility *ef)
 {
-    unsigned int mask;
+    sccb_mask_t mask;
     BusChild *kid;
     SCLPEventClass *child;
 
@@ -77,9 +77,9 @@ static unsigned int get_host_send_mask(SCLPEventFacility *ef)
     return mask;
 }
 
-static unsigned int get_host_receive_mask(SCLPEventFacility *ef)
+static sccb_mask_t get_host_receive_mask(SCLPEventFacility *ef)
 {
-    unsigned int mask;
+    sccb_mask_t mask;
     BusChild *kid;
     SCLPEventClass *child;
 
@@ -189,7 +189,7 @@ out:
 }
 
 static uint16_t handle_sccb_read_events(SCLPEventFacility *ef, SCCB *sccb,
-                                        unsigned int mask)
+                                        sccb_mask_t mask)
 {
     uint16_t rc;
     int slen;
@@ -242,8 +242,8 @@ static void copy_mask(uint8_t *dst, uint8_t *src, uint16_t dst_len,
 
 static void read_event_data(SCLPEventFacility *ef, SCCB *sccb)
 {
-    unsigned int sclp_active_selection_mask;
-    unsigned int sclp_cp_receive_mask;
+    sccb_mask_t sclp_active_selection_mask;
+    sccb_mask_t sclp_cp_receive_mask;
 
     ReadEventData *red = (ReadEventData *) sccb;
 
@@ -285,7 +285,7 @@ static void write_event_mask(SCLPEventFacility *ef, SCCB *sccb)
 {
     WriteEventMask *we_mask = (WriteEventMask *) sccb;
     uint16_t mask_length = be16_to_cpu(we_mask->mask_length);
-    uint32_t tmp_mask;
+    sccb_mask_t tmp_mask;
 
     if (!mask_length || (mask_length > SCLP_EVENT_MASK_LEN_MAX) ||
         ((mask_length != 4) && !ef->allow_all_mask_sizes)) {
diff --git a/hw/s390x/sclpcpu.c b/hw/s390x/sclpcpu.c
index 3ee890b..50c021b 100644
--- a/hw/s390x/sclpcpu.c
+++ b/hw/s390x/sclpcpu.c
@@ -37,12 +37,12 @@ void raise_irq_cpu_hotplug(void)
     sclp_service_interrupt(0);
 }
 
-static unsigned int send_mask(void)
+static sccb_mask_t send_mask(void)
 {
     return SCLP_EVENT_MASK_CONFIG_MGT_DATA;
 }
 
-static unsigned int receive_mask(void)
+static sccb_mask_t receive_mask(void)
 {
     return 0;
 }
diff --git a/hw/s390x/sclpquiesce.c b/hw/s390x/sclpquiesce.c
index 0241643..1c8f5c9 100644
--- a/hw/s390x/sclpquiesce.c
+++ b/hw/s390x/sclpquiesce.c
@@ -28,12 +28,12 @@ static bool can_handle_event(uint8_t type)
     return type == SCLP_EVENT_SIGNAL_QUIESCE;
 }
 
-static unsigned int send_mask(void)
+static sccb_mask_t send_mask(void)
 {
     return SCLP_EVENT_MASK_SIGNAL_QUIESCE;
 }
 
-static unsigned int receive_mask(void)
+static sccb_mask_t receive_mask(void)
 {
     return 0;
 }
diff --git a/include/hw/s390x/event-facility.h b/include/hw/s390x/event-facility.h
index 5119b9b..0e2b761 100644
--- a/include/hw/s390x/event-facility.h
+++ b/include/hw/s390x/event-facility.h
@@ -28,12 +28,14 @@
 #define SCLP_EVENT_SIGNAL_QUIESCE               0x1d
 
 /* SCLP event masks */
-#define SCLP_EVENT_MASK_SIGNAL_QUIESCE          0x00000008
-#define SCLP_EVENT_MASK_MSG_ASCII               0x00000040
-#define SCLP_EVENT_MASK_CONFIG_MGT_DATA         0x10000000
-#define SCLP_EVENT_MASK_OP_CMD                  0x80000000
-#define SCLP_EVENT_MASK_MSG                     0x40000000
-#define SCLP_EVENT_MASK_PMSGCMD                 0x00800000
+#define SCLP_EVMASK(T)  (1ULL << (sizeof(sccb_mask_t) * 8 - (T)))
+
+#define SCLP_EVENT_MASK_OP_CMD          SCLP_EVMASK(SCLP_EVENT_OPRTNS_COMMAND)
+#define SCLP_EVENT_MASK_MSG             SCLP_EVMASK(SCLP_EVENT_MESSAGE)
+#define SCLP_EVENT_MASK_CONFIG_MGT_DATA SCLP_EVMASK(SCLP_EVENT_CONFIG_MGT_DATA)
+#define SCLP_EVENT_MASK_PMSGCMD         SCLP_EVMASK(SCLP_EVENT_PMSGCMD)
+#define SCLP_EVENT_MASK_MSG_ASCII       SCLP_EVMASK(SCLP_EVENT_ASCII_CONSOLE_DATA)
+#define SCLP_EVENT_MASK_SIGNAL_QUIESCE  SCLP_EVMASK(SCLP_EVENT_SIGNAL_QUIESCE)
 
 #define SCLP_UNCONDITIONAL_READ                 0x00
 #define SCLP_SELECTIVE_READ                     0x01
@@ -71,6 +73,8 @@ typedef struct WriteEventMask {
 #define WEM_RECEIVE_MASK(wem, mask_len) ((wem)->masks + 2 * (mask_len))
 #define WEM_SEND_MASK(wem, mask_len) ((wem)->masks + 3 * (mask_len))
 
+typedef uint32_t sccb_mask_t;
+
 typedef struct EventBufferHeader {
     uint16_t length;
     uint8_t  type;
@@ -160,7 +164,7 @@ typedef struct WriteEventData {
 typedef struct ReadEventData {
     SCCBHeader h;
     union {
-        uint32_t mask;
+        sccb_mask_t mask;
         EventBufferHeader ebh;
     };
 } QEMU_PACKED ReadEventData;
@@ -177,10 +181,10 @@ typedef struct SCLPEventClass {
     int (*exit)(SCLPEvent *event);
 
     /* get SCLP's send mask */
-    unsigned int (*get_send_mask)(void);
+    sccb_mask_t (*get_send_mask)(void);
 
     /* get SCLP's receive mask */
-    unsigned int (*get_receive_mask)(void);
+    sccb_mask_t (*get_receive_mask)(void);
 
     int (*read_event_data)(SCLPEvent *event, EventBufferHeader *evt_buf_hdr,
                            int *slen);
-- 
2.7.4

^ permalink raw reply related	[flat|nested] 8+ messages in thread

* [Qemu-devel] [PATCH v2 3/3] s390x/sclp: extend SCLP event masks to 64 bits
  2018-02-22 16:22 [Qemu-devel] [PATCH v2 0/3] s390x/sclp: 64 bit event masks Claudio Imbrenda
  2018-02-22 16:22 ` [Qemu-devel] [PATCH v2 1/3] s390x/sclp: proper support of larger send and receive masks Claudio Imbrenda
  2018-02-22 16:22 ` [Qemu-devel] [PATCH v2 2/3] s390x/sclp: clean up sclp masks Claudio Imbrenda
@ 2018-02-22 16:22 ` Claudio Imbrenda
  2018-02-23 10:37 ` [Qemu-devel] [PATCH v2 0/3] s390x/sclp: 64 bit event masks Cornelia Huck
  3 siblings, 0 replies; 8+ messages in thread
From: Claudio Imbrenda @ 2018-02-22 16:22 UTC (permalink / raw)
  To: cohuck; +Cc: qemu-s390x, qemu-devel, borntraeger

Extend the SCLP event masks to 64 bits.

Notice that using any of the new bits results in a state that cannot be
migrated to an older version.

Signed-off-by: Claudio Imbrenda <imbrenda@linux.vnet.ibm.com>
---
 hw/s390x/event-facility.c         | 43 +++++++++++++++++++++++++++++++++------
 include/hw/s390x/event-facility.h |  2 +-
 2 files changed, 38 insertions(+), 7 deletions(-)

diff --git a/hw/s390x/event-facility.c b/hw/s390x/event-facility.c
index 94fe948..f9e8898 100644
--- a/hw/s390x/event-facility.c
+++ b/hw/s390x/event-facility.c
@@ -30,7 +30,10 @@ struct SCLPEventFacility {
     SysBusDevice parent_obj;
     SCLPEventsBus sbus;
     /* guest's receive mask */
-    sccb_mask_t receive_mask;
+    union {
+        uint32_t receive_mask_compat32;
+        sccb_mask_t receive_mask;
+    };
     /*
      * when false, we keep the same broken, backwards compatible behaviour as
      * before, allowing only masks of size exactly 4; when true, we implement
@@ -262,7 +265,7 @@ static void read_event_data(SCLPEventFacility *ef, SCCB *sccb)
     case SCLP_SELECTIVE_READ:
         copy_mask((uint8_t *)&sclp_active_selection_mask, (uint8_t *)&red->mask,
                   sizeof(sclp_active_selection_mask), ef->mask_length);
-        sclp_active_selection_mask = be32_to_cpu(sclp_active_selection_mask);
+        sclp_active_selection_mask = be64_to_cpu(sclp_active_selection_mask);
         if (!sclp_cp_receive_mask ||
             (sclp_active_selection_mask & ~sclp_cp_receive_mask)) {
             sccb->h.response_code =
@@ -302,13 +305,13 @@ static void write_event_mask(SCLPEventFacility *ef, SCCB *sccb)
     /* keep track of the guest's capability masks */
     copy_mask((uint8_t *)&tmp_mask, WEM_CP_RECEIVE_MASK(we_mask, mask_length),
               sizeof(tmp_mask), mask_length);
-    ef->receive_mask = be32_to_cpu(tmp_mask);
+    ef->receive_mask = be64_to_cpu(tmp_mask);
 
     /* return the SCLP's capability masks to the guest */
-    tmp_mask = cpu_to_be32(get_host_receive_mask(ef));
+    tmp_mask = cpu_to_be64(get_host_receive_mask(ef));
     copy_mask(WEM_RECEIVE_MASK(we_mask, mask_length), (uint8_t *)&tmp_mask,
               mask_length, sizeof(tmp_mask));
-    tmp_mask = cpu_to_be32(get_host_send_mask(ef));
+    tmp_mask = cpu_to_be64(get_host_send_mask(ef));
     copy_mask(WEM_SEND_MASK(we_mask, mask_length), (uint8_t *)&tmp_mask,
               mask_length, sizeof(tmp_mask));
 
@@ -369,6 +372,21 @@ static void command_handler(SCLPEventFacility *ef, SCCB *sccb, uint64_t code)
     }
 }
 
+static bool vmstate_event_facility_mask64_needed(void *opaque)
+{
+    SCLPEventFacility *ef = opaque;
+
+    return (ef->receive_mask & 0xFFFFFFFF) != 0;
+}
+
+static int vmstate_event_facility_mask64_pre_load(void *opaque)
+{
+    SCLPEventFacility *ef = opaque;
+
+    ef->receive_mask &= ~0xFFFFFFFFULL;
+    return 0;
+}
+
 static bool vmstate_event_facility_mask_length_needed(void *opaque)
 {
     SCLPEventFacility *ef = opaque;
@@ -384,6 +402,18 @@ static int vmstate_event_facility_mask_length_pre_load(void *opaque)
     return 0;
 }
 
+static const VMStateDescription vmstate_event_facility_mask64 = {
+    .name = "vmstate-event-facility/mask64",
+    .version_id = 0,
+    .minimum_version_id = 0,
+    .needed = vmstate_event_facility_mask64_needed,
+    .pre_load = vmstate_event_facility_mask64_pre_load,
+    .fields = (VMStateField[]) {
+        VMSTATE_UINT64(receive_mask, SCLPEventFacility),
+        VMSTATE_END_OF_LIST()
+     }
+};
+
 static const VMStateDescription vmstate_event_facility_mask_length = {
     .name = "vmstate-event-facility/mask_length",
     .version_id = 0,
@@ -402,10 +432,11 @@ static const VMStateDescription vmstate_event_facility = {
     .version_id = 0,
     .minimum_version_id = 0,
     .fields = (VMStateField[]) {
-        VMSTATE_UINT32(receive_mask, SCLPEventFacility),
+        VMSTATE_UINT32(receive_mask_compat32, SCLPEventFacility),
         VMSTATE_END_OF_LIST()
      },
     .subsections = (const VMStateDescription * []) {
+        &vmstate_event_facility_mask64,
         &vmstate_event_facility_mask_length,
         NULL
      }
diff --git a/include/hw/s390x/event-facility.h b/include/hw/s390x/event-facility.h
index 0e2b761..06ba4ea 100644
--- a/include/hw/s390x/event-facility.h
+++ b/include/hw/s390x/event-facility.h
@@ -73,7 +73,7 @@ typedef struct WriteEventMask {
 #define WEM_RECEIVE_MASK(wem, mask_len) ((wem)->masks + 2 * (mask_len))
 #define WEM_SEND_MASK(wem, mask_len) ((wem)->masks + 3 * (mask_len))
 
-typedef uint32_t sccb_mask_t;
+typedef uint64_t sccb_mask_t;
 
 typedef struct EventBufferHeader {
     uint16_t length;
-- 
2.7.4

^ permalink raw reply related	[flat|nested] 8+ messages in thread

* Re: [Qemu-devel] [PATCH v2 1/3] s390x/sclp: proper support of larger send and receive masks
  2018-02-22 16:22 ` [Qemu-devel] [PATCH v2 1/3] s390x/sclp: proper support of larger send and receive masks Claudio Imbrenda
@ 2018-02-23 10:31   ` Cornelia Huck
  2018-02-23 11:18     ` Claudio Imbrenda
  0 siblings, 1 reply; 8+ messages in thread
From: Cornelia Huck @ 2018-02-23 10:31 UTC (permalink / raw)
  To: Claudio Imbrenda; +Cc: qemu-s390x, qemu-devel, borntraeger

On Thu, 22 Feb 2018 17:22:57 +0100
Claudio Imbrenda <imbrenda@linux.vnet.ibm.com> wrote:

> Until 67915de9f0383ccf4a ("s390x/event-facility: variable-length event masks")
> we only supported sclp event masks of size exactly 4 bytes, even though

s/of size/of a size of/

> the archiecture allows the guests to set up sclp event masks from 1 to

s/archiecture/architecture/

> 1021 bytes in length.
> After that patch, the behaviour was almost compliant, but some issues
> were still remaining, in particular regarding the handling of selective
> reads and migration.
> 
> When setting the sclp event mask, a mask size is also specified. Until
> now we only considered the size in order to decide which bits to save
> in the internal state. On the other hand, when a guest performs a
> selective read, it sends a mask, but it does not specify a size; the
> implied size is the size of the last mask that has been set.
> 
> Specifying bits in the mask of selective read that are not available in
> the internal mask should return an error, and bits past the end of the
> mask should obviously be ignored. This can only be achieved by keeping
> track of the lenght of the mask.
> 
> The mask length is thus now part of the internal state that needs to be
> migrated.
> 
> This patch fixes the handling of selective reads, whose size will now
> match the length of the event mask, as per architecture.
> 
> While the default behaviour is to be compliant with the architecture,
> when using older machine models the old broken behaviour is selected
> (allowing only masks of size exactly 4), in order to be able to migrate
> toward older versions.
> 
> Fixes: 67915de9f0383ccf4a ("s390x/event-facility: variable-length event masks")
> Signed-off-by: Claudio Imbrenda <imbrenda@linux.vnet.ibm.com>
> ---
>  hw/s390x/event-facility.c  | 91 +++++++++++++++++++++++++++++++++++++++-------
>  hw/s390x/s390-virtio-ccw.c |  8 +++-
>  2 files changed, 85 insertions(+), 14 deletions(-)

Looks reasonable.

^ permalink raw reply	[flat|nested] 8+ messages in thread

* Re: [Qemu-devel] [PATCH v2 0/3] s390x/sclp: 64 bit event masks
  2018-02-22 16:22 [Qemu-devel] [PATCH v2 0/3] s390x/sclp: 64 bit event masks Claudio Imbrenda
                   ` (2 preceding siblings ...)
  2018-02-22 16:22 ` [Qemu-devel] [PATCH v2 3/3] s390x/sclp: extend SCLP event masks to 64 bits Claudio Imbrenda
@ 2018-02-23 10:37 ` Cornelia Huck
  2018-02-23 11:19   ` Claudio Imbrenda
  3 siblings, 1 reply; 8+ messages in thread
From: Cornelia Huck @ 2018-02-23 10:37 UTC (permalink / raw)
  To: Claudio Imbrenda; +Cc: qemu-s390x, qemu-devel, borntraeger

On Thu, 22 Feb 2018 17:22:56 +0100
Claudio Imbrenda <imbrenda@linux.vnet.ibm.com> wrote:

> Until 67915de9f0383ccf4a ("s390x/event-facility: variable-length event masks")
> we only supported 32bit sclp event masks, even though the archiecture
> allows the guests to set up sclp event masks up to 1021 bytes in length.
> With that patch the behaviour was almost compliant, but some issues were
> still remaining, in particular regarding the handling of selective reads
> and migration.
> 
> This patchset fixes migration and the handling of selective reads, and
> puts in place the support for 64-bit sclp event masks internally.
> 
> A new property of the sclp-event device switches between the 32bit masks
> and the compliant behaviour. The property is bound to the machine
> version, so older machines keep the old broken behaviour, allowing for
> migration, but the default is the compliant implementation.
> 
> Fixes: 67915de9f0383ccf4a ("s390x/event-facility: variable-length event masks")
> 
> v1 -> v2 
> * improved comments and patch descriptions to better explain why we need
>   this, including better description of the old broken behaviour
> * rename SCLPEVMSK to SCLP_EVMASK to improve readability
> * removed some unneded variable initializations
> * fixed a pre-existing typo
> 
> Claudio Imbrenda (3):
>   s390x/sclp: proper support of larger send and receive masks
>   s390x/sclp: clean up sclp masks
>   s390x/sclp: extend SCLP event masks to 64 bits
> 
>  hw/char/sclpconsole-lm.c          |   4 +-
>  hw/char/sclpconsole.c             |   4 +-
>  hw/s390x/event-facility.c         | 150 +++++++++++++++++++++++++++++++-------
>  hw/s390x/s390-virtio-ccw.c        |   8 +-
>  hw/s390x/sclpcpu.c                |   4 +-
>  hw/s390x/sclpquiesce.c            |   4 +-
>  include/hw/s390x/event-facility.h |  22 +++---
>  7 files changed, 151 insertions(+), 45 deletions(-)
> 

This looks sane as far as I can see without having access to the
documentation.

BTW, I saw you also extended the masks in the Linux (guest) code (at
least, I see patches queued on the s390 features branch...) You also
might want to adjust the "Linux uses 4 byte masks" comment in the code
in the future.

Before I queue this, I'd like some r-bs/a-bs by folks with access to
the documentation.

^ permalink raw reply	[flat|nested] 8+ messages in thread

* Re: [Qemu-devel] [PATCH v2 1/3] s390x/sclp: proper support of larger send and receive masks
  2018-02-23 10:31   ` Cornelia Huck
@ 2018-02-23 11:18     ` Claudio Imbrenda
  0 siblings, 0 replies; 8+ messages in thread
From: Claudio Imbrenda @ 2018-02-23 11:18 UTC (permalink / raw)
  To: Cornelia Huck; +Cc: qemu-s390x, qemu-devel, borntraeger

On Fri, 23 Feb 2018 11:31:46 +0100
Cornelia Huck <cohuck@redhat.com> wrote:

> On Thu, 22 Feb 2018 17:22:57 +0100
> Claudio Imbrenda <imbrenda@linux.vnet.ibm.com> wrote:
> 
> > Until 67915de9f0383ccf4a ("s390x/event-facility: variable-length
> > event masks") we only supported sclp event masks of size exactly 4
> > bytes, even though  
> 
> s/of size/of a size of/

will fix

> > the archiecture allows the guests to set up sclp event masks from 1
> > to  
> 
> s/archiecture/architecture/

will fix

> > 1021 bytes in length.
> > After that patch, the behaviour was almost compliant, but some
> > issues were still remaining, in particular regarding the handling
> > of selective reads and migration.
> > 
> > When setting the sclp event mask, a mask size is also specified.
> > Until now we only considered the size in order to decide which bits
> > to save in the internal state. On the other hand, when a guest
> > performs a selective read, it sends a mask, but it does not specify
> > a size; the implied size is the size of the last mask that has been
> > set.
> > 
> > Specifying bits in the mask of selective read that are not
> > available in the internal mask should return an error, and bits
> > past the end of the mask should obviously be ignored. This can only
> > be achieved by keeping track of the lenght of the mask.
> > 
> > The mask length is thus now part of the internal state that needs
> > to be migrated.
> > 
> > This patch fixes the handling of selective reads, whose size will
> > now match the length of the event mask, as per architecture.
> > 
> > While the default behaviour is to be compliant with the
> > architecture, when using older machine models the old broken
> > behaviour is selected (allowing only masks of size exactly 4), in
> > order to be able to migrate toward older versions.
> > 
> > Fixes: 67915de9f0383ccf4a ("s390x/event-facility: variable-length
> > event masks") Signed-off-by: Claudio Imbrenda
> > <imbrenda@linux.vnet.ibm.com> ---
> >  hw/s390x/event-facility.c  | 91
> > +++++++++++++++++++++++++++++++++++++++-------
> > hw/s390x/s390-virtio-ccw.c |  8 +++- 2 files changed, 85
> > insertions(+), 14 deletions(-)  
> 
> Looks reasonable.
> 

^ permalink raw reply	[flat|nested] 8+ messages in thread

* Re: [Qemu-devel] [PATCH v2 0/3] s390x/sclp: 64 bit event masks
  2018-02-23 10:37 ` [Qemu-devel] [PATCH v2 0/3] s390x/sclp: 64 bit event masks Cornelia Huck
@ 2018-02-23 11:19   ` Claudio Imbrenda
  0 siblings, 0 replies; 8+ messages in thread
From: Claudio Imbrenda @ 2018-02-23 11:19 UTC (permalink / raw)
  To: Cornelia Huck; +Cc: qemu-s390x, qemu-devel, borntraeger

On Fri, 23 Feb 2018 11:37:55 +0100
Cornelia Huck <cohuck@redhat.com> wrote:

> On Thu, 22 Feb 2018 17:22:56 +0100
> Claudio Imbrenda <imbrenda@linux.vnet.ibm.com> wrote:
> 
> > Until 67915de9f0383ccf4a ("s390x/event-facility: variable-length
> > event masks") we only supported 32bit sclp event masks, even though
> > the archiecture allows the guests to set up sclp event masks up to
> > 1021 bytes in length. With that patch the behaviour was almost
> > compliant, but some issues were still remaining, in particular
> > regarding the handling of selective reads and migration.
> > 
> > This patchset fixes migration and the handling of selective reads,
> > and puts in place the support for 64-bit sclp event masks
> > internally.
> > 
> > A new property of the sclp-event device switches between the 32bit
> > masks and the compliant behaviour. The property is bound to the
> > machine version, so older machines keep the old broken behaviour,
> > allowing for migration, but the default is the compliant
> > implementation.
> > 
> > Fixes: 67915de9f0383ccf4a ("s390x/event-facility: variable-length
> > event masks")
> > 
> > v1 -> v2 
> > * improved comments and patch descriptions to better explain why we
> > need this, including better description of the old broken behaviour
> > * rename SCLPEVMSK to SCLP_EVMASK to improve readability
> > * removed some unneded variable initializations
> > * fixed a pre-existing typo
> > 
> > Claudio Imbrenda (3):
> >   s390x/sclp: proper support of larger send and receive masks
> >   s390x/sclp: clean up sclp masks
> >   s390x/sclp: extend SCLP event masks to 64 bits
> > 
> >  hw/char/sclpconsole-lm.c          |   4 +-
> >  hw/char/sclpconsole.c             |   4 +-
> >  hw/s390x/event-facility.c         | 150
> > +++++++++++++++++++++++++++++++-------
> > hw/s390x/s390-virtio-ccw.c        |   8 +-
> > hw/s390x/sclpcpu.c                |   4 +-
> > hw/s390x/sclpquiesce.c            |   4 +-
> > include/hw/s390x/event-facility.h |  22 +++--- 7 files changed, 151
> > insertions(+), 45 deletions(-) 
> 
> This looks sane as far as I can see without having access to the
> documentation.
> 
> BTW, I saw you also extended the masks in the Linux (guest) code (at
> least, I see patches queued on the s390 features branch...) You also
> might want to adjust the "Linux uses 4 byte masks" comment in the code
> in the future.

will fix, so that I don't have to remember to do stuff in the future :)
 
> Before I queue this, I'd like some r-bs/a-bs by folks with access to
> the documentation.
> 

^ permalink raw reply	[flat|nested] 8+ messages in thread

end of thread, other threads:[~2018-02-23 11:19 UTC | newest]

Thread overview: 8+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-02-22 16:22 [Qemu-devel] [PATCH v2 0/3] s390x/sclp: 64 bit event masks Claudio Imbrenda
2018-02-22 16:22 ` [Qemu-devel] [PATCH v2 1/3] s390x/sclp: proper support of larger send and receive masks Claudio Imbrenda
2018-02-23 10:31   ` Cornelia Huck
2018-02-23 11:18     ` Claudio Imbrenda
2018-02-22 16:22 ` [Qemu-devel] [PATCH v2 2/3] s390x/sclp: clean up sclp masks Claudio Imbrenda
2018-02-22 16:22 ` [Qemu-devel] [PATCH v2 3/3] s390x/sclp: extend SCLP event masks to 64 bits Claudio Imbrenda
2018-02-23 10:37 ` [Qemu-devel] [PATCH v2 0/3] s390x/sclp: 64 bit event masks Cornelia Huck
2018-02-23 11:19   ` Claudio Imbrenda

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.