[v2] spapr, spapr_numa: fix max-associativity-domains

[PATCH v2 3/3] spapr_numa.c: fix ibm, max-associativity-domains calculation

Posted by Daniel Henrique Barboza 5 years ago

The current logic for calculating 'maxdomain' making it a sum of
numa_state->num_nodes with spapr->gpu_numa_id. spapr->gpu_numa_id is
used as a index to determine the next available NUMA id that a
given NVGPU can use.

The problem is that the initial value of gpu_numa_id, for any topology
that has more than one NUMA node, is equal to numa_state->num_nodes.
This means that our maxdomain will always be, at least, twice the
amount of existing NUMA nodes. This means that a guest with 4 NUMA
nodes will end up with the following max-associativity-domains:

rtas/ibm,max-associativity-domains
                 00000004 00000008 00000008 00000008 00000008

This overtuning of maxdomains doesn't go unnoticed in the guest, being
detected in SLUB during boot:

 dmesg | grep SLUB
[    0.000000] SLUB: HWalign=128, Order=0-3, MinObjects=0, CPUs=4, Nodes=8

SLUB is detecting 8 total nodes, with 4 nodes being online.

This patch fixes ibm,max-associativity-domains by considering the amount
of NVGPUs NUMA nodes presented in the guest, instead of just
spapr->gpu_numa_id.

Reported-by: Cédric Le Goater <clg@kaod.org>
Tested-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
---
 hw/ppc/spapr_numa.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/hw/ppc/spapr_numa.c b/hw/ppc/spapr_numa.c
index a757dd88b8..779f18b994 100644
--- a/hw/ppc/spapr_numa.c
+++ b/hw/ppc/spapr_numa.c
@@ -311,6 +311,8 @@ void spapr_numa_write_rtas_dt(SpaprMachineState *spapr, void *fdt, int rtas)
 {
     MachineState *ms = MACHINE(spapr);
     SpaprMachineClass *smc = SPAPR_MACHINE_GET_CLASS(spapr);
+    uint32_t number_nvgpus_nodes = spapr->gpu_numa_id -
+                                   spapr_numa_initial_nvgpu_numa_id(ms);
     uint32_t refpoints[] = {
         cpu_to_be32(0x4),
         cpu_to_be32(0x3),
@@ -318,7 +320,7 @@ void spapr_numa_write_rtas_dt(SpaprMachineState *spapr, void *fdt, int rtas)
         cpu_to_be32(0x1),
     };
     uint32_t nr_refpoints = ARRAY_SIZE(refpoints);
-    uint32_t maxdomain = ms->numa_state->num_nodes + spapr->gpu_numa_id;
+    uint32_t maxdomain = ms->numa_state->num_nodes + number_nvgpus_nodes;
     uint32_t maxdomains[] = {
         cpu_to_be32(4),
         cpu_to_be32(maxdomain),
-- 
2.26.2

Re: [PATCH v2 3/3] spapr_numa.c: fix ibm,max-associativity-domains calculation

Posted by David Gibson 5 years ago

On Thu, Jan 28, 2021 at 02:42:13PM -0300, Daniel Henrique Barboza wrote:
> The current logic for calculating 'maxdomain' making it a sum of
> numa_state->num_nodes with spapr->gpu_numa_id. spapr->gpu_numa_id is
> used as a index to determine the next available NUMA id that a
> given NVGPU can use.
> 
> The problem is that the initial value of gpu_numa_id, for any topology
> that has more than one NUMA node, is equal to numa_state->num_nodes.
> This means that our maxdomain will always be, at least, twice the
> amount of existing NUMA nodes. This means that a guest with 4 NUMA
> nodes will end up with the following max-associativity-domains:
> 
> rtas/ibm,max-associativity-domains
>                  00000004 00000008 00000008 00000008 00000008
> 
> This overtuning of maxdomains doesn't go unnoticed in the guest, being
> detected in SLUB during boot:
> 
>  dmesg | grep SLUB
> [    0.000000] SLUB: HWalign=128, Order=0-3, MinObjects=0, CPUs=4, Nodes=8
> 
> SLUB is detecting 8 total nodes, with 4 nodes being online.
> 
> This patch fixes ibm,max-associativity-domains by considering the amount
> of NVGPUs NUMA nodes presented in the guest, instead of just
> spapr->gpu_numa_id.
> 
> Reported-by: Cédric Le Goater <clg@kaod.org>
> Tested-by: Cédric Le Goater <clg@kaod.org>
> Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

Applied, thanks.

> ---
>  hw/ppc/spapr_numa.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/hw/ppc/spapr_numa.c b/hw/ppc/spapr_numa.c
> index a757dd88b8..779f18b994 100644
> --- a/hw/ppc/spapr_numa.c
> +++ b/hw/ppc/spapr_numa.c
> @@ -311,6 +311,8 @@ void spapr_numa_write_rtas_dt(SpaprMachineState *spapr, void *fdt, int rtas)
>  {
>      MachineState *ms = MACHINE(spapr);
>      SpaprMachineClass *smc = SPAPR_MACHINE_GET_CLASS(spapr);
> +    uint32_t number_nvgpus_nodes = spapr->gpu_numa_id -
> +                                   spapr_numa_initial_nvgpu_numa_id(ms);
>      uint32_t refpoints[] = {
>          cpu_to_be32(0x4),
>          cpu_to_be32(0x3),
> @@ -318,7 +320,7 @@ void spapr_numa_write_rtas_dt(SpaprMachineState *spapr, void *fdt, int rtas)
>          cpu_to_be32(0x1),
>      };
>      uint32_t nr_refpoints = ARRAY_SIZE(refpoints);
> -    uint32_t maxdomain = ms->numa_state->num_nodes + spapr->gpu_numa_id;
> +    uint32_t maxdomain = ms->numa_state->num_nodes + number_nvgpus_nodes;
>      uint32_t maxdomains[] = {
>          cpu_to_be32(4),
>          cpu_to_be32(maxdomain),

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

Re: [PATCH v2 3/3] spapr_numa.c: fix ibm,max-associativity-domains calculation

Posted by Greg Kurz 5 years ago

On Thu, 28 Jan 2021 14:42:13 -0300
Daniel Henrique Barboza <danielhb413@gmail.com> wrote:

> The current logic for calculating 'maxdomain' making it a sum of
> numa_state->num_nodes with spapr->gpu_numa_id. spapr->gpu_numa_id is
> used as a index to determine the next available NUMA id that a
> given NVGPU can use.
> 
> The problem is that the initial value of gpu_numa_id, for any topology
> that has more than one NUMA node, is equal to numa_state->num_nodes.
> This means that our maxdomain will always be, at least, twice the
> amount of existing NUMA nodes. This means that a guest with 4 NUMA
> nodes will end up with the following max-associativity-domains:
> 
> rtas/ibm,max-associativity-domains
>                  00000004 00000008 00000008 00000008 00000008
> 
> This overtuning of maxdomains doesn't go unnoticed in the guest, being
> detected in SLUB during boot:
> 
>  dmesg | grep SLUB
> [    0.000000] SLUB: HWalign=128, Order=0-3, MinObjects=0, CPUs=4, Nodes=8
> 
> SLUB is detecting 8 total nodes, with 4 nodes being online.
> 
> This patch fixes ibm,max-associativity-domains by considering the amount
> of NVGPUs NUMA nodes presented in the guest, instead of just
> spapr->gpu_numa_id.
> 
> Reported-by: Cédric Le Goater <clg@kaod.org>
> Tested-by: Cédric Le Goater <clg@kaod.org>
> Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
> ---

Reviewed-by: Greg Kurz <groug@kaod.org>

>  hw/ppc/spapr_numa.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/hw/ppc/spapr_numa.c b/hw/ppc/spapr_numa.c
> index a757dd88b8..779f18b994 100644
> --- a/hw/ppc/spapr_numa.c
> +++ b/hw/ppc/spapr_numa.c
> @@ -311,6 +311,8 @@ void spapr_numa_write_rtas_dt(SpaprMachineState *spapr, void *fdt, int rtas)
>  {
>      MachineState *ms = MACHINE(spapr);
>      SpaprMachineClass *smc = SPAPR_MACHINE_GET_CLASS(spapr);
> +    uint32_t number_nvgpus_nodes = spapr->gpu_numa_id -
> +                                   spapr_numa_initial_nvgpu_numa_id(ms);
>      uint32_t refpoints[] = {
>          cpu_to_be32(0x4),
>          cpu_to_be32(0x3),
> @@ -318,7 +320,7 @@ void spapr_numa_write_rtas_dt(SpaprMachineState *spapr, void *fdt, int rtas)
>          cpu_to_be32(0x1),
>      };
>      uint32_t nr_refpoints = ARRAY_SIZE(refpoints);
> -    uint32_t maxdomain = ms->numa_state->num_nodes + spapr->gpu_numa_id;
> +    uint32_t maxdomain = ms->numa_state->num_nodes + number_nvgpus_nodes;
>      uint32_t maxdomains[] = {
>          cpu_to_be32(4),
>          cpu_to_be32(maxdomain),