config-6.1.0-rc1-metze.01 e35184f321518acadb681928a016da21a9a20c13 origin/master disabled CONFIG_OVERLAY_FS CONFIG_SHIFT_FS CONFIG_DELL_UART_BACKLIGHT

-CONFIG_OVERLAY_FS=m
-# CONFIG_OVERLAY_FS_REDIRECT_DIR is not set
-CONFIG_OVERLAY_FS_REDIRECT_ALWAYS_FOLLOW=y
-# CONFIG_OVERLAY_FS_INDEX is not set
-CONFIG_OVERLAY_FS_XINO_AUTO=y
-# CONFIG_OVERLAY_FS_METACOPY is not set
+# CONFIG_OVERLAY_FS is not set

-CONFIG_SHIFT_FS=m
-CONFIG_SHIFT_FS_POSIX_ACL=y
+# CONFIG_SHIFT_FS is not set

IORING_SEND_ZC_REPORT_USAGE

io_uring/net: fail zc sendmsg when unsupported by socket

The previous patch fails zerocopy send requests for protocols that don't
support it, do the same for zerocopy sendmsg.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>

io_uring/net: fail zc send when unsupported by socket

If a protocol doesn't support zerocopy it will silently fall back to
copying. This type of behaviour has always been a source of troubles
so it's better to fail such requests instead.

Cc: <stable@vger.kernel.org> # 6.0
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>

net: flag sockets supporting msghdr originated zerocopy

We need an efficient way in io_uring to check whether a socket supports
zerocopy with msghdr provided ubuf_info. Add a new flag into the struct
socket flags fields.

Cc: <stable@vger.kernel.org> # 6.0
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>

config-6.0.0-rc7-metze.01 rebased on v6.0-rc7 + origin/master c3e0e1e23c70455916ff3472072437b3605c6cfe

io_ring_ctx->__spare_list

config-6.0.0-rc6-metze.02 rebase on kernel-dk-block/for-6.1/io_uring

config-6.0.0-rc6-metze.01

config-6.0.0-rc5-metze.08

io_uring: introduce io_uring_probe_op.op_features

This will allow us to specify per operation feature bits,
in order to let userspace figure out addtitional features for
an existing opcode over time.

Signed-off-by: Stefan Metzmacher <metze@samba.org>

.not_supported

net: wire up support for file_operations->uring_cmd()

Pass it through the proto_ops->uring_cmd() handler, so we can plumb it
through all the way to the proto->uring_cmd() handler later
as required.

Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Stefan Metzmacher <metze@samba.org>

TODO io_uring_livepatch io_uring_livepatch_io_nop[_prep]

Build your main kernel with this patch.
Then you should be able to easily modify it.

I typically use a script like this:

> cat build-load-linux-io_uring_livepatch.sh
set -x
rsync -a -P --delete-after /root/linux.git/fs/ /root/tmp-io_uring_livepatch-build/fs/
rsync -a -P --delete-after /root/linux.git/include/ /root/tmp-io_uring_livepatch-build/include/
rsync -a -P --delete-after /root/linux.git/io_uring/ /root/tmp-io_uring_livepatch-build/io_uring/
rsync -a -P --delete-after /root/linux.git/io_uring_livepatch/ /root/tmp-io_uring_livepatch-build/io_uring_livepatch/
pushd /root/tmp-io_uring_livepatch-build/io_uring_livepatch/ && {
        make rmmod
        sleep 1
        rmmod io_uring_livepatch
        sleep 1
        rmmod io_uring_livepatch
        sleep 1
        rmmod io_uring_livepatch
        sleep 1
        rmmod io_uring_livepatch
        sleep 1
        make insmod_unsigned
}

io_uring: add header gards to internal header files

This is not strictly needed, but it makes it easier to
create livepatches in order to prototype io_uring fixes.

Signed-off-by: Stefan Metzmacher <metze@samba.org>

scripts/mod/modpost.c MAX_UNRESOLVED_REPORTS»·10000

io-wq: Fix memory leak in worker creation

If the CPU mask allocation for a node fails, then the memory allocated for
the 'io_wqe' struct of the current node doesn't get freed on the error
handling path, since it has not yet been added to the 'wqes' array.

This was spotted when fuzzing v6.1-rc1 with Syzkaller:
BUG: memory leak
unreferenced object 0xffff8880093d5000 (size 1024):
  comm "syz-executor.2", pid 7701, jiffies 4295048595 (age 13.900s)
  hex dump (first 32 bytes):
    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
  backtrace:
    [<00000000cb463369>] __kmem_cache_alloc_node+0x18e/0x720
    [<00000000147a3f9c>] kmalloc_node_trace+0x2a/0x130
    [<000000004e107011>] io_wq_create+0x7b9/0xdc0
    [<00000000c38b2018>] io_uring_alloc_task_context+0x31e/0x59d
    [<00000000867399da>] __io_uring_add_tctx_node.cold+0x19/0x1ba
    [<000000007e0e7a79>] io_uring_setup.cold+0x1b80/0x1dce
    [<00000000b545e9f6>] __x64_sys_io_uring_setup+0x5d/0x80
    [<000000008a8a7508>] do_syscall_64+0x5d/0x90
    [<000000004ac08bec>] entry_SYSCALL_64_after_hwframe+0x63/0xcd

Fixes: 0e03496d1967 ("io-wq: use private CPU mask")
Cc: stable@vger.kernel.org
Signed-off-by: Rafael Mendonca <rafaelmendsr@gmail.com>
Link: https://lore.kernel.org/r/20221020014710.902201-1-rafaelmendsr@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

io_uring/msg_ring: Fix NULL pointer dereference in io_msg_send_fd()

Syzkaller produced the below call trace:

 BUG: KASAN: null-ptr-deref in io_msg_ring+0x3cb/0x9f0
 Write of size 8 at addr 0000000000000070 by task repro/16399

 CPU: 0 PID: 16399 Comm: repro Not tainted 6.1.0-rc1 #28
 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.11.0-2.el7
 Call Trace:
  <TASK>
  dump_stack_lvl+0xcd/0x134
  ? io_msg_ring+0x3cb/0x9f0
  kasan_report+0xbc/0xf0
  ? io_msg_ring+0x3cb/0x9f0
  kasan_check_range+0x140/0x190
  io_msg_ring+0x3cb/0x9f0
  ? io_msg_ring_prep+0x300/0x300
  io_issue_sqe+0x698/0xca0
  io_submit_sqes+0x92f/0x1c30
  __do_sys_io_uring_enter+0xae4/0x24b0
....
 RIP: 0033:0x7f2eaf8f8289
 RSP: 002b:00007fff40939718 EFLAGS: 00000246 ORIG_RAX: 00000000000001aa
 RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f2eaf8f8289
 RDX: 0000000000000000 RSI: 0000000000006f71 RDI: 0000000000000004
 RBP: 00007fff409397a0 R08: 0000000000000000 R09: 0000000000000039
 R10: 0000000000000000 R11: 0000000000000246 R12: 00000000004006d0
 R13: 00007fff40939880 R14: 0000000000000000 R15: 0000000000000000
  </TASK>
 Kernel panic - not syncing: panic_on_warn set ...

We don't have a NULL check on file_ptr in io_msg_send_fd() function,
so when file_ptr is NUL src_file is also NULL and get_file()
dereferences a NULL pointer and leads to above crash.

Add a NULL check to fix this issue.

Fixes: e6130eba8a84 ("io_uring: add support for passing fixed file descriptors")
Reported-by: syzkaller <syzkaller@googlegroups.com>
Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com>
Link: https://lore.kernel.org/r/20221019171218.1337614-1-harshit.m.mogalapalli@oracle.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

io_uring/rw: remove leftover debug statement

This debug statement was never meant to go into the upstream release,
kill it off before it ends up in a release. It was just part of the
testing for the initial version of the patch.

Fixes: 2ec33a6c3cca ("io_uring/rw: ensure kiocb_end_write() is always called")
Signed-off-by: Jens Axboe <axboe@kernel.dk>

io_uring: don't iopoll from io_ring_ctx_wait_and_kill()

We should not be completing requests from a task context that has already
undergone io_uring cancellations, i.e. __io_uring_cancel(), as there are
some assumptions, e.g. around cached task refs draining. Remove
iopolling from io_ring_ctx_wait_and_kill() as it can be called later
after PF_EXITING is set with the last task_work run.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/7c03cc91455c4a1af49c6b9cbda4e57ea467aa11.1665891182.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

io_uring: reuse io_alloc_req()

Don't duplicate io_alloc_req() in io_req_caches_free() but reuse the
helper.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/6005fc88274864a49fc3096c22d8bdd605cf8576.1665891182.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

io_uring: kill hot path fixed file bitmap debug checks

We test file_table.bitmap in io_file_get_fixed() to check invariants,
don't do it, it's expensive and was showing up in profiles. No reports of
this triggering has come in. Move the check to the file clear instead,
which will still catch any wrong usage.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/cf77f2ded68d2e5b2bc7355784d969837d48e023.1665891182.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

io_uring: remove FFS_SCM

THe lifetime of SCM'ed files is bound to ring_sock, which is destroyed
strictly after we're done with registered file tables. This means there
is no need for the FFS_SCM hack, which was not available on 32-bit builds
anyway.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Link: https://lore.kernel.org/r/984226a1045adf42dc35d8bd7fb5a8bbfa472ce1.1665891182.git.asml.silence@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

config-6.0.0-rc5-metze.07

config-6.0.0-rc5-metze.06

git rm debian/rules

config-6.0.0-rc5-metze.05

config-6.0.0-rc5-metze.04

config-6.0.0-rc5-metze.03

config-6.0.0-rc5-metze.02

zero_fops generic_file_splice_read???

config-6.0.0-rc5-metze.01 CONFIG_LOCALVERSION_AUTO=y

module_param(msg_zerocopy_loopback,...

net/ipv4/tcp.c SPLICE_F_NONBLOCK SPLICE_F_WAITALL

config-6.0.0-1003-oem

$(EXTRAVERSION)-metze.01

UBUNTU: Ubuntu-oem-6.0-6.0.0-1004.4

Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: [Config] update configs and annotations after rebase to 6.0-rc5

Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: link-to-tracker: update tracking bug

BugLink: https://bugs.launchpad.net/bugs/1990391
Properties: no-test-build
Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: Start new release

Ignore: yes
Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: SAUCE: platform/x86/intel: pmc/core: Add Raptor Lake support to pmc core driver

BugLink: https://bugs.launchpad.net/bugs/1988461
Add Raptor Lake client parts (both RPL and RPL_S) support to pmc core
driver. Raptor Lake client parts reuse all the Alder Lake PCH IPs.

Cc: Srinivas Pandruvada <srinivas.pandruvada@intel.com>
Cc: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: David Box <david.e.box@intel.com>
Signed-off-by: Gayatri Kammela <gayatri.kammela@linux.intel.com>
(backported from https://lore.kernel.org/platform-driver-x86/20220830182001.3693030-1-gayatri.kammela@linux.intel.com/raw)
Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: SAUCE: platform/x86: int3472: support independent clock and LED GPIOs

BugLink: https://bugs.launchpad.net/bugs/1989046
(backported from https://github.com/intel/ipu6-drivers/blob/2b8a592dc63e117d8f5af5f32a10d7e0651832d2/patch/int3472-support-independent-clock-and-LED-gpios-5.17+.patch)
Signed-off-by: You-Sheng Yang (vicamo) <vicamo.yang@canonical.com>
Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: [Config] Enable audio for AMD PinkSardine

BugLink: https://bugs.launchpad.net/bugs/1989518
Signed-off-by: Hui Wang <hui.wang@canonical.com>
Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: Ubuntu-oem-6.0-6.0.0-1003.3

Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: link-to-tracker: update tracking bug

BugLink: https://bugs.launchpad.net/bugs/1988821
Properties: no-test-build
Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: [Config] Disable IWLMEI

BugLink: https://bugs.launchpad.net/bugs/1987312
Signed-off-by: You-Sheng Yang (vicamo) <vicamo.yang@canonical.com>
Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: Start new release

Ignore: yes
Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: [Config] Enable CONFIG_AMD_PMF=m.

BugLink: https://bugs.launchpad.net/bugs/1987670
Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: Ubuntu-oem-6.0-6.0.0-1002.2

Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: [Config] update configs after rebase to 6.0-rc3

Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: link-to-tracker: update tracking bug

BugLink: https://bugs.launchpad.net/bugs/1988142
Properties: no-test-build
Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: Start new release

Ignore: yes
Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: Ubuntu-oem-6.0-6.0.0-1001.1

Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: link-to-tracker: update tracking bug

BugLink: https://bugs.launchpad.net/bugs/1987392
Properties: no-test-build
Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: [Config] Enable CONFIG_HSA_AMD_P2P

BugLink: https://bugs.launchpad.net/bugs/1987394
Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: [Config] Update after rebase to unstable.

Ignore: yes
Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: Start new release

Ignore: yes
Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: Packaging changes for oem flavour

Do the other changes, then
sed -i 's/amd64-generic/amd64-oem/' debian.oem/config/annotations
and finally run updateconfigs.

Ignore: yes
Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: Stub oem packaging

cp -r debian.master debian.oem
find debian.oem  \
  -name '*armhf*' \
  -o -name '*arm64*' \
  -o -name '*i386*' \
  -o -name '*ppc64el*' \
  -o -name '*x32*' \
  -o -name '*riscv64*' \
  -o -name '*s390x*' \
  -o -name '*lowlatency*' \
  -o -name '*lpae*' \
  -o -name '*64k*' \
  | xargs rm -r
find debian.oem | xargs rename 's/generic/oem/'
git add debian.oem

Ignore: yes
Signed-off-by: Timo Aaltonen <timo.aaltonen@canonical.com>

UBUNTU: Ubuntu-unstable-6.0.0-8.8

Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: link-to-tracker: update tracking bug

BugLink: https://bugs.launchpad.net/bugs/1990101
Properties: no-test-build
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: Rebase to v6.0-rc6

Ignore: yes
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: Start new release

Ignore: yes
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: Ubuntu-unstable-6.0.0-7.7

Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: [Config] update configs and annotations after rebase to 6.0-rc5

In particular drop CONFIG_ARM64_BTI_KERNEL that is not supported anymore
if the kernel is built with gcc, see:

 c0a454b9044f ("arm64/bti: Disable in kernel BTI when cross section thunks are broken")

Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: [Packaging[ get rid of unnecessary 'set -x'

Drop a 'set -x' previously added for debugging purposes by commit
7339f2ad00dc ("UBUNTU: [Packaging]: always use fully qualified <triplet>-gcc-VER toolchain")

Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: debian/dkms-versions -- update from kernel-versions (main/master)

BugLink: https://bugs.launchpad.net/bugs/1786013
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: link-to-tracker: update tracking bug

BugLink: https://bugs.launchpad.net/bugs/1989331
Properties: no-test-build
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: Rebase to v6.0-rc5

Ignore: yes
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: Start new release

Ignore: yes
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: [Packaging] resync getabis

BugLink: https://bugs.launchpad.net/bugs/1786013
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: debian/dkms-versions -- update from kernel-versions (main/master)

BugLink: https://bugs.launchpad.net/bugs/1786013
Signed-off-by: Dimitri John Ledkov <dimitri.ledkov@canonical.com>

UBUNTU: [Packaging] Pass kernel build_arch to dkms

This enables dkms-build script to cross-compile dkms modules, for
example zfs during cross-build.

Signed-off-by: Dimitri John Ledkov <dimitri.ledkov@canonical.com>

UBUNTU: debian/dkms-versions -- update from kernel-versions (main/master)

BugLink: https://bugs.launchpad.net/bugs/1786013
Signed-off-by: Dimitri John Ledkov <dimitri.ledkov@canonical.com>

UBUNTU: [Config] Switch from DECOMP_SINGLE to DECOMP_MULTI_PERCPU

BugLink: https://bugs.launchpad.net/bugs/1980861
Overall MULTI_PERCPU has similar or better performance on the wide
range of instance types and squashfs filesystems.

Signed-off-by: Dimitri John Ledkov <dimitri.ledkov@canonical.com>
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: Ubuntu-unstable-6.0.0-6.6

Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: link-to-tracker: update tracking bug

BugLink: https://bugs.launchpad.net/bugs/1988706
Properties: no-test-build
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: Rebase to v6.0-rc4

Ignore: yes
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: Start new release

Ignore: yes
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: [Packaging] fix URLs used to retrieve kernel ABI

Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: [Config] update toolchain versions

Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: [Config] enable security-related configs

Make sure that the following config options are enabled to provide an
additional level of security:

 - CONFIG_ZERO_CALL_USED_REGS: automatically zero local registers on
   function exit

 - CONFIG_INIT_STACK_ALL_ZERO: automatically initialize variables
   defined in the stack to zero

 - CONFIG_SHADOW_CALL_STACK: enables compiler's Shadow Call Stack (see
   https://gcc.gnu.org/onlinedocs/gcc/Instrumentation-Options.html#Instrumentation-Options)

Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: [Packaging] temporarily disable signing for ppc64el and s390x

Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: SAUCE: selftests: tdx: Test GetReport TDX attestation feature

BugLink: https://bugs.launchpad.net/bugs/1988120
In TDX guest, attestation is used to verify the trustworthiness of a
TD. During the TD bring-up, Intel TDX module measures and records the
initial contents and configuration of TD, and at runtime, TD software
uses runtime measurement registers (RMTRs) to measure and record
details related to kernel image, command line params, ACPI tables,
initrd, etc. At TD runtime, Intel SGX attestation infrastructure is
re-used to attest to these measurement data.

First step in the TDX attestation process is to get the TDREPORT data.
It is a fixed size data structure generated by the TDX module which
includes the above mentioned measurements data, a MAC to protect the
integerity of the TDREPORT, and a 64-Byte of user specified data passed
during TDREPORT request which can uniquely identify the TDREPORT.

Intel's TDX guest driver exposes TDX_CMD_GET_REPORT IOCTL interface to
get the TDREPORT from the user space.

Add a kernel selftest module to test this ABI and verify the validity
of generated TDREPORT.

Reviewed-by: Tony Luck <tony.luck@intel.com>
Reviewed-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com>
(cherry picked from https://lore.kernel.org/lkml/YwZbcpvmJt50YDa%2F@zn.tnic/T/#mcab45a43d062ba98e2c768c9667d965e34ee64e7)
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: SAUCE: x86/tdx: Add TDX Guest attestation interface driver

BugLink: https://bugs.launchpad.net/bugs/1988120
In TDX guest, attestation is used to verify the trustworthiness of a TD
to other entities before provisioning secrets to the TD. Such attestation
process is required by 3rd party servers before sending sensitive
information to TD guests. One usage example is to get encryption keys
from the key server for mounting the encrypted rootfs or secondary drive.

During the TD launch, the initial contents and configuration of the TD
are recorded by the Intel TDX module in build time measurement register
(MRTD). It is a SHA384 digest created using data from TD private pages(
including TD firmware) and the configuration of the TD.

After TD build, run-time measurement registers (RTMRs)  can be used by
the guest TD software to extend the TD measurements. TDX supports 4
RTMR registers, and TDG.MR.RTMR.EXTEND TDCALL is used to update the
RTMR registers securely. RTMRs are mainly used to record measurements
related to sections like the kernel image, command line parameters,
initrd, ACPI tables, firmware data, configuration firmware volume (CFV)
of TDVF, etc. For more details, please refer to TDX Virtual Firmware
design specification, sec titled "TD Measurement".

At TD runtime, the Intel TDX module reuses the Intel SGX attestation
infrastructure to provide support for attesting to these measurements
as described below.

The attestation process consists of two steps: TDREPORT generation and
Quote generation.

TDREPORT (TDREPORT_STRUCT) is a fixed-size data structure generated by
the TDX module which contains TD-specific information (such as TD
measurements), platform security version, and the MAC to protect the
integrity of the TDREPORT. The TD kernel uses TDCALL[TDG.MR.REPORT] to
get the TDREPORT from the TDX module. A user-provided 64-Byte
REPORTDATA is used as input and included in the TDREPORT. Typically it
can be some nonce provided by attestation service so the TDREPORT can
be verified uniquely. More details about TDREPORT can be found in
Intel TDX Module specification, section titled "TDG.MR.REPORT Leaf".

TDREPORT by design can only be verified on local platform as the MAC
key is bound to the platform. To support remote verification of the
TDREPORT, TDX leverages Intel SGX Quote Enclave (QE) to verify the
TDREPORT locally and convert it to a remote verifiable Quote.

After getting the TDREPORT, the second step of the attestation process
is to send it to the QE to generate the Quote. TDX doesn't support SGX
inside the TD, so the QE can be deployed in the host, or in another
legacy VM with SGX support. QE uses EVERIFYREPORT2 instruction to help
check the integrity of TDREPORT and if it is valid, a certified quote
signing key is used to sign the Quote. How to send the TDREPORT to QE
and receive the Quote is implementation and deployment specific.

Implement a basic TD guest misc driver to allow TD userspace to get the
TDREPORT. The TD userspace attestation software can get the TDREPORT
and then choose whatever communication channel available (i.e. vsock
or hypercall) to send the TDREPORT to QE and receive the Quote.

Also note that explicit access permissions are not enforced in this
driver because the quote and measurements are not a secret. However
the access permissions of the device node can be used to set any
desired access policy. The udev default is usually root access
only.

Operations like getting TDREPORT or Quote generation involves sending
a blob of data as input and getting another blob of data as output. It
was considered to use a sysfs interface for this, but it doesn't fit
well into the standard sysfs model for configuring values. It would be
possible to do read/write on files, but it would need multiple file
descriptors, which would be somewhat messy. IOCTLs seems to be the best
fitting and simplest model for this use case. This is similar to AMD
SEV platform, which also uses IOCTL interface to support attestation.

Any distribution enabling TDX is also expected to need attestation. So
enable it by default with TDX guest support.

Reviewed-by: Tony Luck <tony.luck@intel.com>
Reviewed-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Kai Huang <kai.huang@intel.com>
Acked-by: Wander Lairson Costa <wander@redhat.com>
Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com>
(cherry picked from https://lore.kernel.org/lkml/YwZbcpvmJt50YDa%2F@zn.tnic/T/#mcab45a43d062ba98e2c768c9667d965e34ee64e7)
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: Ubuntu-unstable-6.0.0-5.5

Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: [Config] update configs after rebase to 6.0-rc3

Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: link-to-tracker: update tracking bug

BugLink: https://bugs.launchpad.net/bugs/1988019
Properties: no-test-build
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: Rebase to v6.0-rc3

Ignore: yes
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: Start new release

Ignore: yes
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: [Packaging] Remove circular dependency between image & modules

Currently we have circular dependencies between
linux-image-[unsigned]-ABI and linux-modules[-extra] packages.

Debian policy point of view this is bad and does lead to worse
dependency resolution cycles in apt upon kernel installs and upgrades,
but mostly harmless as apt has learned to deal with that a long time
ago.

It is harmful for kernels that have additional types of debs or snaps
that deliver vmlinux.

In addition to regular linux-image[-unsigned] we produce various FDE
linux-images like things with builtin initrds, i.e. linux-fde,
linux-uc22, FIT images etc. All of them vendor in vmlinuz, and do not
need to have linux-image[-unsigned] installed. They also often do not
want to pull in any of the bootloaders listed as recommends by the
linux-image-ABI package either, as they use alternative methods to
boot (i.e. gadget snap, nullboot, flashkernel, weird cloud firmware
provided stuff, etc).

Remove linux-modules[-extra] -> linux-image[-unsigned] dependency. Add
linux-modules-extra -> linux-modules dependency (because extra modules
use symbols from the regular modules package).

This will enable us to build clean meta-packages that pull in
appropriate image, fde, uc22, FIT kernel image with matching modules,
without installing redudant copies of vmlinux or bootloaders.

Signed-off-by: Dimitri John Ledkov <dimitri.ledkov@canonical.com>
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: [Config] Set CONFIG_MTD_RAW_NAND=m for riscv64

BugLink: https://bugs.launchpad.net/bugs/1981437
These modules are built for all other architectures but s390x.
Let's build them for riscv64 too.

Signed-off-by: Emil Renner Berthing <emil.renner.berthing@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Dimitri John Ledkov <dimitri.ledkov@canonical.com>
Signed-off-by: Dimitri John Ledkov <dimitri.ledkov@canonical.com>

UBUNTU: [Config] Enable CONFIG_DTPM for riscv64

BugLink: https://bugs.launchpad.net/bugs/1981437
Enable power capping for the dynamic thermal power management for
riscv64. This option is not annotated, but let's enable it just like
all the other architectures.

Signed-off-by: Emil Renner Berthing <emil.renner.berthing@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Dimitri John Ledkov <dimitri.ledkov@canonical.com>
Signed-off-by: Dimitri John Ledkov <dimitri.ledkov@canonical.com>

UBUNTU: [Config] Enable CONFIG_DRM_RCAR_USE_LVDS for riscv64

BugLink: https://bugs.launchpad.net/bugs/1981437
This option is not annotated, but let's enable this feature for riscv64
like all the other architectures have.

Signed-off-by: Emil Renner Berthing <emil.renner.berthing@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Dimitri John Ledkov <dimitri.ledkov@canonical.com>
Signed-off-by: Dimitri John Ledkov <dimitri.ledkov@canonical.com>

UBUNTU: [Config] Align riscv64 CONFIG_LSM with other architectures

BugLink: https://bugs.launchpad.net/bugs/1981437
The CONFIG_LSM option for riscv64 is doesn't enable the landlock module
at boot. Add it and align riscv64 with all the other archituctures.

Signed-off-by: Emil Renner Berthing <emil.renner.berthing@canonical.com>
Acked-by: Tim Gardner <tim.gardner@canonical.com>
Acked-by: Dimitri John Ledkov <dimitri.ledkov@canonical.com>
Signed-off-by: Dimitri John Ledkov <dimitri.ledkov@canonical.com>

UBUNTU: [Packaging] use unversioned default gcc

Whilst cbd/dkms/etc cannot yet dynamically install desired
build-dependencies, simply use the default gcc compiler without an
explicit features.

Signed-off-by: Dimitri John Ledkov <dimitri.ledkov@canonical.com>
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: [Packaging]: always use fully qualified <triplet>-gcc-VER toolchain

Always define $(gcc) variable and export it. Always use fully
qualified <triplet>-gcc-VER toolchains for native (GNU build) and
cross (GNU host) compilation. Generate matching build-dependencies for
native and cross builds.

This ensures that native and cross builds use the same fully qualified
toolchain binary, fixed in time, irrespective of gcc-defaults updates.

Signed-off-by: Dimitri John Ledkov <dimitri.ledkov@canonical.com>
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: Ubuntu-unstable-6.0.0-4.4

Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: link-to-tracker: update tracking bug

BugLink: https://bugs.launchpad.net/bugs/1987517
Properties: no-test-build
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>

UBUNTU: Start new release

Ignore: yes
Signed-off-by: Andrea Righi <andrea.righi@canonical.com>