Merge branch 'riscv-cpu_map_topo' of git://git.kernel.org/pub/scm/linux/kernel/git...
authorPalmer Dabbelt <palmer@rivosinc.com>
Thu, 14 Jul 2022 21:40:10 +0000 (14:40 -0700)
committerPalmer Dabbelt <palmer@rivosinc.com>
Thu, 14 Jul 2022 21:40:10 +0000 (14:40 -0700)
commiteabfbd92a67827a3d899a9dda4d840e75ff11309
tree46acd9b47bc7faa1f0c9d64ba5e5691089771f8d
parenta8914b6cb8dd0e8ba215e6047f8ed0c88ba4ea7b
parentd9d193dea8666bbf69fc21c5bdcdabaa34a466e3
Merge branch 'riscv-cpu_map_topo' of git://git./linux/kernel/git/palmer/linux.git into for-next

RISC-V: Add cpu-map topology information nodes

It was reported to me that the Hive Unmatched incorrectly reports
its topology to hwloc, but the StarFive VisionFive did in [0] &
a subsequent off-list email from Brice (the hwloc maintainer).
This turned out not to be entirely true, the /downstream/ version
of the VisionFive does work correctly but not upstream, as the
downstream devicetree has a cpu-map node that was added recently.

This series adds a cpu-map node to all upstream devicetrees, which
I have tested on mpfs & fu540. The first patch is lifted directly
from the downstream StarFive devicetree.

0: https://github.com/open-mpi/hwloc/issues/536

Link: https://lore.kernel.org/linux-riscv/20220705190435.1790466-1-mail@conchuod.ie/
[Palmer: except the Microchip DT, that went in via the previous PR.]

* 'riscv-cpu_map_topo' of git://git.kernel.org/pub/scm/linux/kernel/git/palmer/linux.git:
  riscv: dts: canaan: Add k210 topology information
  riscv: dts: sifive: Add fu740 topology information
  riscv: dts: sifive: Add fu540 topology information
  riscv: dts: starfive: Add JH7100 CPU topology