~mreed8855/ubuntu/+source/linux/+git/lunar:config_numa_emu_lunar

Last commit made on 2023-04-13
Get this branch:
git clone -b config_numa_emu_lunar https://git.launchpad.net/~mreed8855/ubuntu/+source/linux/+git/lunar
Only Michael Reed can upload to this branch. If you are Michael Reed please log in for upload directions.

Branch merges

Branch information

Name:
config_numa_emu_lunar
Repository:
lp:~mreed8855/ubuntu/+source/linux/+git/lunar

Recent commits

b1d54d0... by Michael Reed

UBUNTU: [Config] Intel Sapphire Rapids HBM support needs CONFIG_NUMA_EMU

BugLink: https://bugs.launchpad.net/bugs/2008745

Currently Ubuntu kernel has this kernel config disabled.
But in some cases, Intel's Sapphire Rapids High Bandwith
Memory (SPR-HBM) needs this option.

Memory bandwidth has been a bottleneck of increasingly memory bound
workloads. Sapphire Rapids plus HBM is specifically targeted to
cater to these workloads, traditionally served using overprovisioning
of memory devices.

Signed-off-by: Michael Reed <email address hidden>

d33d832... by Andrea Righi

UBUNTU: [Packaging] get rid of unnecessary artifacts in linux-headers

There is no sane reason to ship .o and .cmd artifacts in linux-headers,
so simply drop them (this allows to save ~25MB in the latest
linux-headers-generic).

Moreover, strip Rust's libmacros.so shipped in linux-headers to reduce
the size even more.

Signed-off-by: Andrea Righi <email address hidden>

119fdad... by Andrea Righi

UBUNTU: SAUCE: selftests/bpf: avoid conflicting data types in profiler.inc.h

In profiler.inc.h we define some structs that may conflict with the
names that bpftool emits when it generates vmlinux.h from
/sys/kernel/btf/vmlinux.

Make sure these local data types don't conflict with those generated by
bpftool.

Signed-off-by: Andrea Righi <email address hidden>

bfa22d2... by Andrea Righi

UBUNTU: SAUCE: selftests/bpf: ignore pointer types check with clang

Building bpf selftests with clang can trigger errors like the following:

  CLNG-BPF [test_maps] bpf_iter_netlink.bpf.o
progs/bpf_iter_netlink.c:32:4: error: incompatible pointer types assigning to 'struct sock *' from 'struct sock___17 *' [-Werror,-Wincompatible-pointer-types]
        s = &nlk->sk;
          ^ ~~~~~~~~
1 error generated.

This is due to the fact that bpftool emits duplicate data types with
different names in vmlinux.h (i.e., `struct sock` in this case) and
these types, despite having a different name, represent in fact the same
object.

Add -Wno-incompatible-pointer-types to CLANG_CLAGS to prevent these
errors.

(cherry picked from https://<email address hidden>/T/#u)
Signed-off-by: Andrea Righi <email address hidden>

6d1d37b... by Andrea Righi

Revert "UBUNTU: SAUCE: Revert "efi: random: fix NULL-deref when refreshing seed""

This reverts commit ceed77e05e7b308b0dae06602bb3b986827a5d4e.

Not needed anymore with the folllwing applied:
 a6d8a9c1e5fa ("arm64: efi: Use SMBIOS processor version to key off Ampere quirk")

Signed-off-by: Andrea Righi <email address hidden>

ea0d5fc... by Andrea Righi

Revert "UBUNTU: SAUCE: Revert "efi: random: refresh non-volatile random seed when RNG is initialized""

This reverts commit 5812f3712ae95960725c11214e20f8dc84ce8276.

Not needed anymore with the folllwing applied:
 a6d8a9c1e5fa ("arm64: efi: Use SMBIOS processor version to key off Ampere quirk")

Signed-off-by: Andrea Righi <email address hidden>

7576840... by Ard Biesheuvel <email address hidden>

efi/libstub: smbios: Drop unused 'recsize' parameter

BugLink: https://bugs.launchpad.net/bugs/2015741

We no longer use the recsize argument for locating the string table in
an SMBIOS record, so we can drop it from the internal API.

Signed-off-by: Ard Biesheuvel <email address hidden>
Signed-off-by: Andrea Righi <email address hidden>

77074f5... by Ard Biesheuvel <email address hidden>

arm64: efi: Use SMBIOS processor version to key off Ampere quirk

BugLink: https://bugs.launchpad.net/bugs/2015741

Instead of using the SMBIOS type 1 record 'family' field, which is often
modified by OEMs, use the type 4 'processor ID' and 'processor version'
fields, which are set to a small set of probe-able values on all known
Ampere EFI systems in the field.

Fixes: 550b33cfd4452968 ("arm64: efi: Force the use of ...")
Tested-by: Andrea Righi <email address hidden>
Signed-off-by: Ard Biesheuvel <email address hidden>
(cherry picked from commit eb684408f3ea4856639675d6465f0024e498e4b1)
Signed-off-by: Andrea Righi <email address hidden>

f915c2d... by Ard Biesheuvel <email address hidden>

efi/libstub: smbios: Use length member instead of record struct size

BugLink: https://bugs.launchpad.net/bugs/2015741

The type 1 SMBIOS record happens to always be the same size, but there
are other record types which have been augmented over time, and so we
should really use the length field in the header to decide where the
string table starts.

Fixes: 550b33cfd4452968 ("arm64: efi: Force the use of ...")
Signed-off-by: Ard Biesheuvel <email address hidden>
Signed-off-by: Andrea Righi <email address hidden>

672adac... by Andrea Righi

UBUNTU: Ubuntu-6.2.0-20.20

Signed-off-by: Andrea Righi <email address hidden>