qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Eduardo Habkost <ehabkost@redhat.com>
To: qemu-devel@nongnu.org
Cc: "Igor Mammedov" <imammedo@redhat.com>,
	"Andreas Färber" <afaerber@suse.de>,
	"Gleb Natapov" <gleb@redhat.com>
Subject: [Qemu-devel] [RFC 15/15] generate APIC IDs according to CPU topology (v2)
Date: Tue,  7 Aug 2012 16:56:53 -0300	[thread overview]
Message-ID: <1344369413-9053-16-git-send-email-ehabkost@redhat.com> (raw)
In-Reply-To: <1344369413-9053-1-git-send-email-ehabkost@redhat.com>

This keeps compatibility on machine-types pc-1.1 and lower, and prints a
warning in case the requested configuration won't get the correct
topology.

Changes v1 -> v2:
 - Move code to cpu.c
 - keep using cpu_index on *-user
 - Use SMP.contiguous_apic_ids global property
 - Prints warning in case the compatibility mode will expose incorrect
   topology

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
---
 hw/pc_piix.c      |  4 ++++
 target-i386/cpu.c | 34 ++++++++++++++++++++++++++++++----
 2 files changed, 34 insertions(+), 4 deletions(-)

diff --git a/hw/pc_piix.c b/hw/pc_piix.c
index 0c0096f..f073916 100644
--- a/hw/pc_piix.c
+++ b/hw/pc_piix.c
@@ -375,6 +375,10 @@ static QEMUMachine pc_machine_v1_2 = {
             .driver   = "qxl",\
             .property = "vgamem_mb",\
             .value    = stringify(8),\
+        },{\
+            .driver   = "SMP",\
+            .property = "contiguous_apic_ids",\
+            .value    = "true",\
         }
 
 static QEMUMachine pc_machine_v1_1 = {
diff --git a/target-i386/cpu.c b/target-i386/cpu.c
index 1703373..ea6c7a7 100644
--- a/target-i386/cpu.c
+++ b/target-i386/cpu.c
@@ -23,9 +23,11 @@
 
 #include "cpu.h"
 #include "kvm.h"
+#include "topology.h"
 
 #ifndef CONFIG_USER_ONLY
-#include "sysemu.h"
+#include "hw/qdev.h"
+#include "cpus.h"
 #endif
 
 #include "qemu-option.h"
@@ -492,13 +494,37 @@ static x86_def_t builtin_x86_defs[] = {
     },
 };
 
+#ifdef CONFIG_USER_ONLY
 unsigned int apic_id_for_cpu(int cpu_index)
 {
-    /* right now APIC ID == CPU index. this will eventually change to use
-     * the CPU topology configuration properly
-     */
+    /* *-user doesn't have any CPU topology settings, just use the CPU index */
     return cpu_index;
 }
+#else
+unsigned int apic_id_for_cpu(int cpu_index)
+{
+    const char *contig;
+    bool is_contiguous;
+    unsigned int correct_id;
+    static bool warned = false;
+
+    /* Global property SMP.contiguous_apic_ids=true will keep compatibility
+     * with the old (broken) behavior when calculating APIC IDs
+     */
+    contig = qemu_global_get("SMP", "contiguous_apic_ids");
+    is_contiguous = contig && !strcmp(contig, "true");
+    correct_id = topo_apicid_for_cpu(smp_cores, smp_threads, cpu_index);
+    if (is_contiguous) {
+        if (cpu_index != correct_id && !warned) {
+            fprintf(stderr, "warning: CPU topology in compatibility mode, it will not match the requested topology\n");
+            warned = true;
+        }
+        return cpu_index;
+    } else {
+        return correct_id;
+    }
+}
+#endif
 
 static int cpu_x86_fill_model_id(char *str)
 {
-- 
1.7.11.2

      parent reply	other threads:[~2012-08-07 19:56 UTC|newest]

Thread overview: 25+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-08-07 19:56 [Qemu-devel] [RFC 00/15] attempt to fix CPU topology info on CPU APIC IDs Eduardo Habkost
2012-08-07 19:56 ` [Qemu-devel] [RFC 01/15] cpus.h: include cpu-common.h Eduardo Habkost
2012-08-13 19:06   ` Igor Mammedov
2012-08-07 19:56 ` [Qemu-devel] [RFC 02/15] hw/apic.c: rename bit functions to not conflict with bitops.h (v2) Eduardo Habkost
2012-08-13 19:08   ` Igor Mammedov
2012-08-07 19:56 ` [Qemu-devel] [RFC 03/15] kvm: set vcpu_id to APIC ID instead of CPU index Eduardo Habkost
2012-08-13 19:16   ` Igor Mammedov
2012-08-13 19:59     ` Eduardo Habkost
2012-08-07 19:56 ` [Qemu-devel] [RFC 04/15] i386: create apic_id_for_cpu() function (v2) Eduardo Habkost
2012-08-07 19:56 ` [Qemu-devel] [RFC 05/15] remove FW_CFG_MAX_CPUS from fw_cfg_init() Eduardo Habkost
2012-08-07 19:56 ` [Qemu-devel] [RFC 06/15] pc: set FW_CFG data based on APIC ID calculation Eduardo Habkost
2012-08-13 19:52   ` Igor Mammedov
2012-08-13 20:01     ` Eduardo Habkost
2012-08-07 19:56 ` [Qemu-devel] [RFC 07/15] qdev: allow qdev_prop_parse() to report errors Eduardo Habkost
2012-08-07 19:56 ` [Qemu-devel] [RFC 08/15] move global properties code to global-properties.c Eduardo Habkost
2012-08-07 19:56 ` [Qemu-devel] [RFC 09/15] isolate qdev-independent parts of qdev_prop_set_globals() Eduardo Habkost
2012-08-07 19:56 ` [Qemu-devel] [RFC 10/15] create object_prop_set_globals() Eduardo Habkost
2012-08-07 19:56 ` [Qemu-devel] [RFC 11/15] rename qdev_prop_register_global_list to qemu_globals_register_list Eduardo Habkost
2012-08-07 19:56 ` [Qemu-devel] [RFC 12/15] create qemu_global_get() function Eduardo Habkost
2012-08-07 19:56 ` [Qemu-devel] [RFC 13/15] tests: support target-specific unit tests Eduardo Habkost
2012-08-07 19:56 ` [Qemu-devel] [RFC 14/15] i386: topology & APIC ID utility functions (v2) Eduardo Habkost
2012-08-08 18:57   ` Blue Swirl
2012-08-08 19:06     ` Eduardo Habkost
2012-08-14 17:03   ` Igor Mammedov
2012-08-07 19:56 ` Eduardo Habkost [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1344369413-9053-16-git-send-email-ehabkost@redhat.com \
    --to=ehabkost@redhat.com \
    --cc=afaerber@suse.de \
    --cc=gleb@redhat.com \
    --cc=imammedo@redhat.com \
    --cc=qemu-devel@nongnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).