Certain SMP topologies stress, e.g. 1 thread/core, 2048 cores and
1 socket, stress the current maximum size of the pSeries FDT:

Calling ibm,client-architecture-support...qemu-system-ppc64: error
creating device tree: (fdt_setprop(fdt, offset,
"ibm,processor-segment-sizes", segs, sizeof(segs))): FDT_ERR_NOSPACE

2048 is the default NR_CPUS value for the pSeries kernel. It's expected
that users will want QEMU to be able to handle this kind of
configuration.

Bumping FDT_MAX_SIZE to 2MB is enough for these setups to be created.

Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
---
 include/hw/ppc/spapr.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/include/hw/ppc/spapr.h b/include/hw/ppc/spapr.h
index bf7cab7a2c..3deb382678 100644
--- a/include/hw/ppc/spapr.h
+++ b/include/hw/ppc/spapr.h
@@ -95,7 +95,7 @@ typedef enum {
 #define SPAPR_CAP_FIXED_CCD             0x03
 #define SPAPR_CAP_FIXED_NA              0x10 /* Lets leave a bit of a gap... */
 
-#define FDT_MAX_SIZE                    0x100000
+#define FDT_MAX_SIZE                    0x200000
 
 /*
  * NUMA related macros. MAX_DISTANCE_REF_POINTS was taken
-- 
2.30.2

On Thu, Apr 08, 2021 at 05:40:49PM -0300, Daniel Henrique Barboza wrote:
> Certain SMP topologies stress, e.g. 1 thread/core, 2048 cores and
> 1 socket, stress the current maximum size of the pSeries FDT:
> 
> Calling ibm,client-architecture-support...qemu-system-ppc64: error
> creating device tree: (fdt_setprop(fdt, offset,
> "ibm,processor-segment-sizes", segs, sizeof(segs))): FDT_ERR_NOSPACE
> 
> 2048 is the default NR_CPUS value for the pSeries kernel. It's expected
> that users will want QEMU to be able to handle this kind of
> configuration.
> 
> Bumping FDT_MAX_SIZE to 2MB is enough for these setups to be created.
> 
> Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>

Applied to ppc-for-6.1, thanks.

> ---
>  include/hw/ppc/spapr.h | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/include/hw/ppc/spapr.h b/include/hw/ppc/spapr.h
> index bf7cab7a2c..3deb382678 100644
> --- a/include/hw/ppc/spapr.h
> +++ b/include/hw/ppc/spapr.h
> @@ -95,7 +95,7 @@ typedef enum {
>  #define SPAPR_CAP_FIXED_CCD             0x03
>  #define SPAPR_CAP_FIXED_NA              0x10 /* Lets leave a bit of a gap... */
>  
> -#define FDT_MAX_SIZE                    0x100000
> +#define FDT_MAX_SIZE                    0x200000
>  
>  /*
>   * NUMA related macros. MAX_DISTANCE_REF_POINTS was taken

-- 
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

On 4/8/21 10:40 PM, Daniel Henrique Barboza wrote:
> Certain SMP topologies stress, e.g. 1 thread/core, 2048 cores and
> 1 socket, stress the current maximum size of the pSeries FDT:
> 
> Calling ibm,client-architecture-support...qemu-system-ppc64: error
> creating device tree: (fdt_setprop(fdt, offset,
> "ibm,processor-segment-sizes", segs, sizeof(segs))): FDT_ERR_NOSPACE
> 
> 2048 is the default NR_CPUS value for the pSeries kernel. It's expected
> that users will want QEMU to be able to handle this kind of
> configuration.
> 
> Bumping FDT_MAX_SIZE to 2MB is enough for these setups to be created.
> 
> Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
> ---
>  include/hw/ppc/spapr.h | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/include/hw/ppc/spapr.h b/include/hw/ppc/spapr.h
> index bf7cab7a2c..3deb382678 100644
> --- a/include/hw/ppc/spapr.h
> +++ b/include/hw/ppc/spapr.h
> @@ -95,7 +95,7 @@ typedef enum {
>  #define SPAPR_CAP_FIXED_CCD             0x03
>  #define SPAPR_CAP_FIXED_NA              0x10 /* Lets leave a bit of a gap... */
>  
> -#define FDT_MAX_SIZE                    0x100000
> +#define FDT_MAX_SIZE                    0x200000
>  
>  /*
>   * NUMA related macros. MAX_DISTANCE_REF_POINTS was taken
> 

FYI,

On a very large system, I also had to bump the FDT_MAX_SIZE value in 
softmmu/device-tree.c because QEMU is parsing the host DT. 

Thanks,

C.