Patchwork [RFC,v3,09/19] pc: Add dimm paravirt SRAT info

login
register
mail settings
Submitter Vasilis Liaskovitis
Date Sept. 21, 2012, 11:17 a.m.
Message ID <1348226255-4226-10-git-send-email-vasilis.liaskovitis@profitbricks.com>
Download mbox | patch
Permalink /patch/185705/
State New
Headers show

Comments

Vasilis Liaskovitis - Sept. 21, 2012, 11:17 a.m.
The numa_fw_cfg paravirt interface is extended to include SRAT information for
all hotplug-able dimms. There are 3 words for each hotplug-able memory slot,
denoting start address, size and node proximity. The new info is appended after
existing numa info, so that the fw_cfg layout does not break.  This information
is used by Seabios to build hotplug memory device objects at runtime.
nb_numa_nodes is set to 1 by default (not 0), so that we always pass srat info
to SeaBIOS.

v1->v2:
Dimm SRAT info (#dimms) is appended at end of existing numa fw_cfg in order not
to break existing layout
Documentation of the new fwcfg layout is included in docs/specs/fwcfg.txt

Signed-off-by: Vasilis Liaskovitis <vasilis.liaskovitis@profitbricks.com>
---
 docs/specs/fwcfg.txt |   28 ++++++++++++++++++++++++++++
 hw/pc.c              |   14 ++++++++++++--
 2 files changed, 40 insertions(+), 2 deletions(-)
 create mode 100644 docs/specs/fwcfg.txt
Wen Congyang - Sept. 27, 2012, 3:55 a.m.
At 09/21/2012 07:17 PM, Vasilis Liaskovitis Wrote:
> The numa_fw_cfg paravirt interface is extended to include SRAT information for
> all hotplug-able dimms. There are 3 words for each hotplug-able memory slot,
> denoting start address, size and node proximity. The new info is appended after
> existing numa info, so that the fw_cfg layout does not break.  This information
> is used by Seabios to build hotplug memory device objects at runtime.
> nb_numa_nodes is set to 1 by default (not 0), so that we always pass srat info
> to SeaBIOS.

You forgot to set nb_numa_nodes to 1...

Thanks
Wen Congyang

> 
> v1->v2:
> Dimm SRAT info (#dimms) is appended at end of existing numa fw_cfg in order not
> to break existing layout
> Documentation of the new fwcfg layout is included in docs/specs/fwcfg.txt
> 
> Signed-off-by: Vasilis Liaskovitis <vasilis.liaskovitis@profitbricks.com>
> ---
>  docs/specs/fwcfg.txt |   28 ++++++++++++++++++++++++++++
>  hw/pc.c              |   14 ++++++++++++--
>  2 files changed, 40 insertions(+), 2 deletions(-)
>  create mode 100644 docs/specs/fwcfg.txt
> 
> diff --git a/docs/specs/fwcfg.txt b/docs/specs/fwcfg.txt
> new file mode 100644
> index 0000000..55f96d9
> --- /dev/null
> +++ b/docs/specs/fwcfg.txt
> @@ -0,0 +1,28 @@
> +QEMU<->BIOS Paravirt Documentation
> +--------------------------------------
> +
> +This document describes paravirt data structures passed from QEMU to BIOS.
> +
> +FW_CFG_NUMA paravirt info
> +--------------------
> +The SRAT info passed from QEMU to BIOS has the following layout:
> +
> +-----------------------------------------------------------------------------------------------
> +#nodes | cpu0_pxm | cpu1_pxm | ... | cpulast_pxm | node0_mem | node1_mem | ... | nodelast_mem
> +
> +-----------------------------------------------------------------------------------------------
> +#dimms | dimm0_start | dimm0_sz | dimm0_pxm | ... | dimmlast_start | dimmlast_sz | dimmlast_pxm
> +
> +Entry 0 contains the number of numa nodes (nb_numa_nodes).
> +
> +Entries 1..max_cpus: The next max_cpus entries describe node proximity for each
> +one of the vCPUs in the system.
> +
> +Entries max_cpus+1..max_cpus+nb_numa_nodes+1:  The next nb_numa_nodes entries
> +describe the memory size for each one of the NUMA nodes in the system.
> +
> +Entry max_cpus+nb_numa_nodes+1 contains the number of memory dimms (nb_hp_dimms)
> +
> +The last 3 * nb_hp_dimms entries are organized in triplets: Each triplet contains
> +the physical address offset, size (in bytes), and node proximity for the
> +respective dimm.
> diff --git a/hw/pc.c b/hw/pc.c
> index 2c9664d..f2604ae 100644
> --- a/hw/pc.c
> +++ b/hw/pc.c
> @@ -598,6 +598,7 @@ static void *bochs_bios_init(void)
>      uint8_t *smbios_table;
>      size_t smbios_len;
>      uint64_t *numa_fw_cfg;
> +    uint64_t *hp_dimms_fw_cfg;
>      int i, j;
>  
>      register_ioport_write(0x400, 1, 2, bochs_bios_write, NULL);
> @@ -632,8 +633,10 @@ static void *bochs_bios_init(void)
>      /* allocate memory for the NUMA channel: one (64bit) word for the number
>       * of nodes, one word for each VCPU->node and one word for each node to
>       * hold the amount of memory.
> +     * Finally one word for the number of hotplug memory slots and three words
> +     * for each hotplug memory slot (start address, size and node proximity).
>       */
> -    numa_fw_cfg = g_malloc0((1 + max_cpus + nb_numa_nodes) * 8);
> +    numa_fw_cfg = g_malloc0((2 + max_cpus + nb_numa_nodes + 3 * nb_hp_dimms) * 8);
>      numa_fw_cfg[0] = cpu_to_le64(nb_numa_nodes);
>      for (i = 0; i < max_cpus; i++) {
>          for (j = 0; j < nb_numa_nodes; j++) {
> @@ -646,8 +649,15 @@ static void *bochs_bios_init(void)
>      for (i = 0; i < nb_numa_nodes; i++) {
>          numa_fw_cfg[max_cpus + 1 + i] = cpu_to_le64(node_mem[i]);
>      }
> +
> +    numa_fw_cfg[1 + max_cpus + nb_numa_nodes] = cpu_to_le64(nb_hp_dimms);
> +
> +    hp_dimms_fw_cfg = numa_fw_cfg + 2 + max_cpus + nb_numa_nodes;
> +    if (nb_hp_dimms)
> +        setup_fwcfg_hp_dimms(hp_dimms_fw_cfg);
> +
>      fw_cfg_add_bytes(fw_cfg, FW_CFG_NUMA, (uint8_t *)numa_fw_cfg,
> -                     (1 + max_cpus + nb_numa_nodes) * 8);
> +                     (2 + max_cpus + nb_numa_nodes + 3 * nb_hp_dimms) * 8);
>  
>      return fw_cfg;
>  }

Patch

diff --git a/docs/specs/fwcfg.txt b/docs/specs/fwcfg.txt
new file mode 100644
index 0000000..55f96d9
--- /dev/null
+++ b/docs/specs/fwcfg.txt
@@ -0,0 +1,28 @@ 
+QEMU<->BIOS Paravirt Documentation
+--------------------------------------
+
+This document describes paravirt data structures passed from QEMU to BIOS.
+
+FW_CFG_NUMA paravirt info
+--------------------
+The SRAT info passed from QEMU to BIOS has the following layout:
+
+-----------------------------------------------------------------------------------------------
+#nodes | cpu0_pxm | cpu1_pxm | ... | cpulast_pxm | node0_mem | node1_mem | ... | nodelast_mem
+
+-----------------------------------------------------------------------------------------------
+#dimms | dimm0_start | dimm0_sz | dimm0_pxm | ... | dimmlast_start | dimmlast_sz | dimmlast_pxm
+
+Entry 0 contains the number of numa nodes (nb_numa_nodes).
+
+Entries 1..max_cpus: The next max_cpus entries describe node proximity for each
+one of the vCPUs in the system.
+
+Entries max_cpus+1..max_cpus+nb_numa_nodes+1:  The next nb_numa_nodes entries
+describe the memory size for each one of the NUMA nodes in the system.
+
+Entry max_cpus+nb_numa_nodes+1 contains the number of memory dimms (nb_hp_dimms)
+
+The last 3 * nb_hp_dimms entries are organized in triplets: Each triplet contains
+the physical address offset, size (in bytes), and node proximity for the
+respective dimm.
diff --git a/hw/pc.c b/hw/pc.c
index 2c9664d..f2604ae 100644
--- a/hw/pc.c
+++ b/hw/pc.c
@@ -598,6 +598,7 @@  static void *bochs_bios_init(void)
     uint8_t *smbios_table;
     size_t smbios_len;
     uint64_t *numa_fw_cfg;
+    uint64_t *hp_dimms_fw_cfg;
     int i, j;
 
     register_ioport_write(0x400, 1, 2, bochs_bios_write, NULL);
@@ -632,8 +633,10 @@  static void *bochs_bios_init(void)
     /* allocate memory for the NUMA channel: one (64bit) word for the number
      * of nodes, one word for each VCPU->node and one word for each node to
      * hold the amount of memory.
+     * Finally one word for the number of hotplug memory slots and three words
+     * for each hotplug memory slot (start address, size and node proximity).
      */
-    numa_fw_cfg = g_malloc0((1 + max_cpus + nb_numa_nodes) * 8);
+    numa_fw_cfg = g_malloc0((2 + max_cpus + nb_numa_nodes + 3 * nb_hp_dimms) * 8);
     numa_fw_cfg[0] = cpu_to_le64(nb_numa_nodes);
     for (i = 0; i < max_cpus; i++) {
         for (j = 0; j < nb_numa_nodes; j++) {
@@ -646,8 +649,15 @@  static void *bochs_bios_init(void)
     for (i = 0; i < nb_numa_nodes; i++) {
         numa_fw_cfg[max_cpus + 1 + i] = cpu_to_le64(node_mem[i]);
     }
+
+    numa_fw_cfg[1 + max_cpus + nb_numa_nodes] = cpu_to_le64(nb_hp_dimms);
+
+    hp_dimms_fw_cfg = numa_fw_cfg + 2 + max_cpus + nb_numa_nodes;
+    if (nb_hp_dimms)
+        setup_fwcfg_hp_dimms(hp_dimms_fw_cfg);
+
     fw_cfg_add_bytes(fw_cfg, FW_CFG_NUMA, (uint8_t *)numa_fw_cfg,
-                     (1 + max_cpus + nb_numa_nodes) * 8);
+                     (2 + max_cpus + nb_numa_nodes + 3 * nb_hp_dimms) * 8);
 
     return fw_cfg;
 }