[RFC,v4,0/7] Specifying cache topology on ARM

Message

Alireza Sanaee Dec. 16, 2024, 5:54 p.m. UTC

Specifying the cache layout in virtual machines is useful for
applications and operating systems to fetch accurate information about
the cache structure and make appropriate adjustments. Enforcing correct
sharing information can lead to better optimizations. This patch enables
the specification of cache layout through a command line parameter,
building on a patch set by Intel [1,2]. It uses this set as a
foundation.  The device tree and ACPI/PPTT table, and device tree are
populated based on user-provided information and CPU topology.

Example:


+----------------+                            +----------------+
|    Socket 0    |                            |    Socket 1    |
|    (L3 Cache)  |                            |    (L3 Cache)  |
+--------+-------+                            +--------+-------+
         |                                             |
+--------+--------+                            +--------+--------+
|   Cluster 0     |                            |   Cluster 0     |
|   (L2 Cache)    |                            |   (L2 Cache)    |
+--------+--------+                            +--------+--------+
         |                                             |
+--------+--------+  +--------+--------+    +--------+--------+  +--------+----+
|   Core 0         | |   Core 1        |    |   Core 0        |  |   Core 1    |
|   (L1i, L1d)     | |   (L1i, L1d)    |    |   (L1i, L1d)    |  |   (L1i, L1d)|
+--------+--------+  +--------+--------+    +--------+--------+  +--------+----+
         |                   |                       |                   |
+--------+              +--------+              +--------+          +--------+
|Thread 0|              |Thread 1|              |Thread 1|          |Thread 0|
+--------+              +--------+              +--------+          +--------+
|Thread 1|              |Thread 0|              |Thread 0|          |Thread 1|
+--------+              +--------+              +--------+          +--------+


The following command will represent the system relying on **ACPI PPTT tables**.

./qemu-system-aarch64 \
 -machine virt,smp-cache.0.cache=l1i,smp-cache.0.topology=core,smp-cache.1.cache=l1d,smp-cache.1.topology=core,smp-cache.2.cache=l2,smp-cache.2.topology=cluseter,smp-cache.3.cache=l3,smp-cache.3.topology=socket \
 -cpu max \
 -m 2048 \
 -smp sockets=2,clusters=1,cores=2,threads=2 \
 -kernel ./Image.gz \
 -append "console=ttyAMA0 root=/dev/ram rdinit=/init acpi=force" \
 -initrd rootfs.cpio.gz \
 -bios ./edk2-aarch64-code.fd \
 -nographic

The following command will represent the system relying on **the device tree**.

./qemu-system-aarch64 \
 -machine virt,smp-cache.0.cache=l1i,smp-cache.0.topology=core,smp-cache.1.cache=l1d,smp-cache.1.topology=core,smp-cache.2.cache=l2,smp-cache.2.topology=cluseter,smp-cache.3.cache=l3,smp-cache.3.topology=socket \
 -cpu max \
 -m 2048 \
 -smp sockets=2,clusters=1,cores=2,threads=2 \
 -kernel ./Image.gz \
 -append "console=ttyAMA0 root=/dev/ram rdinit=/init acpi=force" \
 -initrd rootfs.cpio.gz \
 -bios ./edk2-aarch64-code.fd \
 -nographic

Failure cases:
    1) there are cases where QEMU might not have any clusters selected in the
    -smp option, while user specifies caches to be shared at cluster level. In
    this situations, qemu returns error.

    2) There are other scenarios where caches exist in systems' registers but
    not left unspecified by users. In this case qemu returns failure.

    3) At the moment, the device tree is not able to describe caches shared at
    core level. In another word, SMT threads cannot share caches. This will need
    adjustments in the SPEC. It is worth noting that this particular case is completely
    OK in ACPI PPTT tables.

Currently only three levels of caches are supported to be specified from
the command line. However, increasing the value does not require
significant changes. Further, this patch assumes l2 and l3 unified
caches and does not allow l(2/3)(i/d). The level terminology is
thread/core/cluster/socket right now.

Here is the hierarchy assumed in this patch:
Socket level = Cluster level + 1 = Core level + 2 = Thread level + 3;

TODO:
1) Making the code to work with arbitrary levels
2) Separated data and instruction cache at L2 and L3.
3) Additional cache controls.  e.g. size of L3 may not want to just
match the underlying system, because only some of the associated host
CPUs may be bound to this VM.

Depends-on: target/arm/tcg: refine cache descriptions with a wrapper
Depends-on: Msg-id: [e-mail]

Depends-on: Building PPTT with root node and identical implementation flag
Depends-on: Msg-id: [e-mail]

[1] https://lore.kernel.org/kvm/[e-mail]/
[2] https://lore.kernel.org/qemu-devel/[e-mail]/

v3->v4:
Device tree added.

Alireza Sanaee (7):
  i386/cpu: add IsDefined flag to smp-cache property
  target/arm/tcg: increase cache level for cpu=max
  arm/virt.c: add cache hierarchy to device tree
  bios-tables-test: prepare to change ARM ACPI virt PPTT
  hw/acpi/aml-build.c: add cache hierarchy to pptt table
  tests/qtest/bios-table-test: testing new ARM ACPI PPTT topology
  Update the ACPI tables according to the acpi aml_build change, also
    empty bios-tables-test-allowed-diff.h.

 hw/acpi/aml-build.c                        | 235 ++++++++++++-
 hw/arm/virt-acpi-build.c                   |   8 +-
 hw/arm/virt.c                              | 390 +++++++++++++++++++++
 hw/core/machine-smp.c                      |   2 +
 hw/cpu/core.c                              |  97 +++++
 include/hw/acpi/aml-build.h                |   4 +-
 include/hw/arm/virt.h                      |   5 +
 include/hw/boards.h                        |   1 +
 include/hw/cpu/core.h                      |  27 ++
 target/arm/tcg/cpu64.c                     |  13 +
 tests/data/acpi/aarch64/virt/PPTT.topology | Bin 356 -> 540 bytes
 tests/qtest/bios-tables-test.c             |   6 +-
 12 files changed, 779 insertions(+), 9 deletions(-)