qemu-s390x
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [PATCH v14 02/11] s390x/cpu topology: add topology entries on CPU ho


From: Pierre Morel
Subject: Re: [PATCH v14 02/11] s390x/cpu topology: add topology entries on CPU hotplug
Date: Tue, 17 Jan 2023 14:55:50 +0100
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101 Thunderbird/102.5.0



On 1/13/23 19:15, Nina Schoetterl-Glausch wrote:
On Thu, 2023-01-05 at 15:53 +0100, Pierre Morel wrote:
The topology information are attributes of the CPU and are
specified during the CPU device creation.

On hot plug, we gather the topology information on the core,
creates a list of topology entries, each entry contains a single
core mask of each core with identical topology and finaly we
s/finaly/finally/

thx

orders the list in topological order.
s/orders/order/

thx

The topological order is, from higher to lower priority:
- physical topology
     - drawer
     - book
     - socket
     - core origin, offset in 64bit increment from core 0.
- modifier attributes
     - CPU type
     - polarization entitlement
     - dedication

The possibility to insert a CPU in a mask is dependent on the
number of cores allowed in a socket, a book or a drawer, the
checking is done during the hot plug of the CPU to have an
immediate answer.

If the complete topology is not specified, the core is added
in the physical topology based on its core ID and it gets
defaults values for the modifier attributes.

This way, starting QEMU without specifying the topology can
still get some adventage of the CPU topology.

Signed-off-by: Pierre Morel <pmorel@linux.ibm.com>
---
  include/hw/s390x/cpu-topology.h |  48 ++++++
  hw/s390x/cpu-topology.c         | 293 ++++++++++++++++++++++++++++++++
  hw/s390x/s390-virtio-ccw.c      |  10 ++
  hw/s390x/meson.build            |   1 +
  4 files changed, 352 insertions(+)
  create mode 100644 hw/s390x/cpu-topology.c

diff --git a/include/hw/s390x/cpu-topology.h b/include/hw/s390x/cpu-topology.h
index d945b57fc3..b3fd752d8d 100644
--- a/include/hw/s390x/cpu-topology.h
+++ b/include/hw/s390x/cpu-topology.h
@@ -10,7 +10,11 @@
  #ifndef HW_S390X_CPU_TOPOLOGY_H
  #define HW_S390X_CPU_TOPOLOGY_H
+#include "qemu/queue.h"
+#include "hw/boards.h"
+
  #define S390_TOPOLOGY_CPU_IFL   0x03
+#define S390_TOPOLOGY_MAX_ORIGIN ((63 + S390_MAX_CPUS) / 64)
#define S390_TOPOLOGY_POLARITY_HORIZONTAL 0x00
  #define S390_TOPOLOGY_POLARITY_VERTICAL_LOW    0x01
@@ -20,4 +24,48 @@
  #define S390_TOPOLOGY_SHARED    0x00
  #define S390_TOPOLOGY_DEDICATED 0x01
+typedef union s390_topology_id {
+    uint64_t id;
+    struct {
+        uint64_t level_6:8; /* byte 0 BE */
+        uint64_t level_5:8; /* byte 1 BE */
+        uint64_t drawer:8;  /* byte 2 BE */
+        uint64_t book:8;    /* byte 3 BE */
+        uint64_t socket:8;  /* byte 4 BE */
+        uint64_t rsrv:5;
+        uint64_t d:1;
+        uint64_t p:2;       /* byte 5 BE */
+        uint64_t type:8;    /* byte 6 BE */
+        uint64_t origin:2;

This is two bits because it's the core divided by 64, and we have 248 cores at 
most?
Where is this set?

right, must be set so does the core offset in the mask.


+        uint64_t core:6;    /* byte 7 BE */
+    };
+} s390_topology_id;

This struct seems to do double duty, 1. it represents a cpu and 2. a topology 
entry.
You also use it for sorting.
I would suggest to just use a cpu object when referring to a specific cpu and
put the relevant fields directly into the topology entry.

Yes, I can remove the core:6.
After Thomas comment I will change all the bit field for uint8_t.
I think we should not use the real topology entry here if we want to use TGE in the future.

You get rid of the bit field that way.
You'd then need a comparison function for a cpu object and a topology entry.
As long as that isn't the only type pair that shouldn't be too ugly.

+#define TOPO_CPU_MASK       0x000000000000003fUL
+
+typedef struct S390TopologyEntry {
+    s390_topology_id id;
+    QTAILQ_ENTRY(S390TopologyEntry) next;
+    uint64_t mask;
+} S390TopologyEntry;
+
+typedef struct S390Topology {
+    QTAILQ_HEAD(, S390TopologyEntry) list;
+    uint8_t *sockets;
+    CpuTopology *smp;
+} S390Topology;
+
+#ifdef CONFIG_KVM
+bool s390_has_topology(void);
+void s390_topology_set_cpu(MachineState *ms, S390CPU *cpu, Error **errp);
+#else
+static inline bool s390_has_topology(void)
+{
+       return false;
+}
+static inline void s390_topology_set_cpu(MachineState *ms,
+                                         S390CPU *cpu,
+                                         Error **errp) {}
+#endif
+extern S390Topology s390_topology;
+
  #endif
diff --git a/hw/s390x/cpu-topology.c b/hw/s390x/cpu-topology.c
new file mode 100644
index 0000000000..438055c612
--- /dev/null
+++ b/hw/s390x/cpu-topology.c
@@ -0,0 +1,293 @@
+/*
+ * CPU Topology
+ *
+ * Copyright IBM Corp. 2022
+ * Author(s): Pierre Morel <pmorel@linux.ibm.com>
+
+ * This work is licensed under the terms of the GNU GPL, version 2 or (at
+ * your option) any later version. See the COPYING file in the top-level
+ * directory.
+ */
+
+#include "qemu/osdep.h"
+#include "qapi/error.h"
+#include "qemu/error-report.h"
+#include "hw/qdev-properties.h"
+#include "hw/boards.h"
+#include "qemu/typedefs.h"
+#include "target/s390x/cpu.h"
+#include "hw/s390x/s390-virtio-ccw.h"
+#include "hw/s390x/cpu-topology.h"
+
+/*
+ * s390_topology is used to keep the topology information.
+ * .list: queue the topology entries inside which
+ *        we keep the information on the CPU topology.
+ *
+ * .smp: keeps track of the machine topology.
+ *
+ * .socket: tracks information on the count of cores per socket.
+ *
+ */
+S390Topology s390_topology = {
+    .list = QTAILQ_HEAD_INITIALIZER(s390_topology.list),
+    .sockets = NULL, /* will be initialized after the cpu model is realized */

I guess you should do the same for .smp then also.

OK


+};
+
+/**
+ * s390_socket_nb:
+ * @id: s390_topology_id
+ *
+ * Returns the socket number used inside the socket array.
+ */
+static int s390_socket_nb(s390_topology_id id)
+{
+    return (id.socket + 1) * (id.book + 1) * (id.drawer + 1);

This calculation doesn't make a whole lot of sense to me.
It's symmetric with regards to the variables, so (s=0 b=1 d=1)
will have the same result as (s=1 b=0 d=1).
You want the "global" socket number right?
So that would be (drawer * books_per_drawer + book) * sockets_per_book + socket.

yes, already changed


+}
+
+/**
+ * s390_has_topology:
+ *
+ * Return value: if the topology is supported by the machine.
+ */
+bool s390_has_topology(void)
+{
+    return false;
+}
+
+/**
+ * s390_topology_init:
+ * @ms: the machine state where the machine topology is defined
+ *
+ * Keep track of the machine topology.
+ * Allocate an array to keep the count of cores per socket.
+ * The index of the array starts at socket 0 from book 0 and
+ * drawer 0 up to the maximum allowed by the machine topology.
+ */
+static void s390_topology_init(MachineState *ms)
+{
+    CpuTopology *smp = &ms->smp;
+
+    s390_topology.smp = smp;
+    if (!s390_topology.sockets) {

Is this function being called multiple times, or why the if?
Use an assert instead?

I forgot to remove this test.
The caller already check this.


+        s390_topology.sockets = g_new0(uint8_t, smp->sockets *
+                                       smp->books * smp->drawers);
+    }
+}
+
+/**
+ * s390_topology_from_cpu:
+ * @cpu: The S390CPU
+ *
+ * Initialize the topology id from the CPU environment.
+ */
+static s390_topology_id s390_topology_from_cpu(S390CPU *cpu)
+{
+    s390_topology_id topology_id;
+
+    topology_id.core = cpu->env.core_id;
+    topology_id.type = cpu->env.cpu_type;
+    topology_id.p = cpu->env.polarity;
+    topology_id.d = cpu->env.dedicated;
+    topology_id.socket = cpu->env.socket_id;
+    topology_id.book = cpu->env.book_id;
+    topology_id.drawer = cpu->env.drawer_id;
+
+    return topology_id;
+}
+
+/**
+ * s390_topology_set_entry:
+ * @entry: Topology entry to setup
+ * @id: topology id to use for the setup
+ *
+ * Set the core bit inside the topology mask and
+ * increments the number of cores for the socket.
+ */
+static void s390_topology_set_entry(S390TopologyEntry *entry,

Not sure if I like the name, what it does is to add a cpu to the entry.

s390_topology_add_cpu_to_entry() ?




+                                    s390_topology_id id)
+{
+    set_bit(63 - id.core, &entry->mask);

You need to subtract the origin first or that might be negative.

yes, origin is not handled correctly


+    s390_topology.sockets[s390_socket_nb(id)]++;
+}
+
+/**
+ * s390_topology_new_entry:
+ * @id: s390_topology_id to add
+ *
+ * Allocate a new entry and initialize it.
+ *
+ * returns the newly allocated entry.
+ */
+static S390TopologyEntry *s390_topology_new_entry(s390_topology_id id)
+{
+    S390TopologyEntry *entry;
+
+    entry = g_malloc0(sizeof(S390TopologyEntry));
+    entry->id.id = id.id & ~TOPO_CPU_MASK;
+    s390_topology_set_entry(entry, id);
+
+    return entry;
+}
+
+/**
+ * s390_topology_insert:
+ *
+ * @id: s390_topology_id to insert.
+ *
+ * Parse the topology list to find if the entry already
+ * exist and add the core in it.
+ * If it does not exist, allocate a new entry and insert
+ * it in the queue from lower id to greater id.
+ */
+static void s390_topology_insert(s390_topology_id id)
+{
+    S390TopologyEntry *entry;
+    S390TopologyEntry *tmp = NULL;
+    uint64_t new_id;
+
+    new_id = id.id & ~TOPO_CPU_MASK;
+
+    /* First CPU to add to an entry */
+    if (QTAILQ_EMPTY(&s390_topology.list)) {
+        entry = s390_topology_new_entry(id);
+        QTAILQ_INSERT_HEAD(&s390_topology.list, entry, next);
+        return;
+    }
+
+    QTAILQ_FOREACH(tmp, &s390_topology.list, next) {
+        if (new_id == tmp->id.id) {
+            s390_topology_set_entry(tmp, id);
+            return;
+        } else if (new_id < tmp->id.id) {
+            entry = s390_topology_new_entry(id);
+            QTAILQ_INSERT_BEFORE(tmp, entry, next);
+            return;
+        }
+    }
+
+    entry = s390_topology_new_entry(id);
+    QTAILQ_INSERT_TAIL(&s390_topology.list, entry, next);

Consider adding a sentinel entry "at infinity", then that whole code
would simplify.

looks good, thanks.



+}
+
+/**
+ * s390_topology_check:
+ * @errp: Error pointer
+ * id: s390_topology_id to be verified
+ *
+ * The function checks if the topology id fits inside the
+ * system topology.
+ */
+static void s390_topology_check(Error **errp, s390_topology_id id)
+{
+    CpuTopology *smp = s390_topology.smp;
+
+    if (id.socket > smp->sockets) {
+            error_setg(errp, "Unavailable socket: %d", id.socket);
+            return;
+    }
+    if (id.book > smp->books) {
+            error_setg(errp, "Unavailable book: %d", id.book);
+            return;
+    }
+    if (id.drawer > smp->drawers) {
+            error_setg(errp, "Unavailable drawer: %d", id.drawer);
+            return;
+    }
+    if (id.type != S390_TOPOLOGY_CPU_IFL) {
+            error_setg(errp, "Unknown cpu type: %d", id.type);
+            return;
+    }
+    /* Polarity and dedication can never be wrong */
+}
+
+/**
+ * s390_topology_cpu_default:
+ * @errp: Error pointer
+ * @cpu: pointer to a S390CPU
+ *
+ * Setup the default topology for unset attributes.
+ *
+ * The function accept only all all default values or all set values
+ * for the geometry topology.
+ *
+ * The function calculates the (drawer_id, book_id, socket_id)
+ * topology by filling the cores starting from the first socket
+ * (0, 0, 0) up to the last (smp->drawers, smp->books, smp->sockets).
+ *
+ */
+static void s390_topology_cpu_default(Error **errp, S390CPU *cpu)
+{
+    CpuTopology *smp = s390_topology.smp;
+    CPUS390XState *env = &cpu->env;
+
+    /* All geometry topology attributes must be set or all unset */
+    if ((env->socket_id < 0 || env->book_id < 0 || env->drawer_id < 0) &&
+        (env->socket_id >= 0 || env->book_id >= 0 || env->drawer_id >= 0)) {
+        error_setg(errp,
+                   "Please define all or none of the topology geometry 
attributes");
+        return;
+    }
+
+    /* Check if one of the geometry topology is unset */
+    if (env->socket_id < 0) {
+        /* Calculate default geometry topology attributes */
+        env->socket_id = (env->core_id / smp->cores) % smp->sockets;
+        env->book_id = (env->core_id / (smp->sockets * smp->cores)) %
+                       smp->books;
+        env->drawer_id = (env->core_id /
+                          (smp->books * smp->sockets * smp->cores)) %
+                         smp->drawers;
+    }
+}
+
+/**
+ * s390_topology_set_cpu:
+ * @ms: MachineState used to initialize the topology structure on
+ *      first call.
+ * @cpu: the new S390CPU to insert in the topology structure
+ * @errp: the error pointer
+ *
+ * Called from CPU Hotplug to check and setup the CPU attributes
+ * before to insert the CPU in the topology.
+ */
+void s390_topology_set_cpu(MachineState *ms, S390CPU *cpu, Error **errp)
+{
+    Error *local_error = NULL;

Can't you just use ERRP_GUARD ?

I do not think it is necessary and I find it obfuscating.
So, should I?


+    s390_topology_id id;
+
+    /*
+     * We do not want to initialize the topology if the cpu model
+     * does not support topology consequently, we have to wait for

", consequently," I think. Could you do the initialization some where else,
after you know what the cpu model is? Not that I object to doing it this way.


I did not find a better place, it must be done after the CPU model is initialize and before the first CPU is created.
The cpu model is initialized during the early creation of the first cpu.

Any idea?

Thanks.

Regards,
Pierre

--
Pierre Morel
IBM Lab Boeblingen



reply via email to

[Prev in Thread] Current Thread [Next in Thread]