Here is the change log:

```
* SECURITY ISSUE - Fixed a bug where Cipher.update_into accepted Python buffer protocol objects, but allowed immutable buffers [CVE-2023-23931]
* Updated Windows, macOS, and Linux wheels to be compiled with OpenSSL 3.0.8.
```

Change-Id: I3a5f88506c6bc560c66c6b0bb8005bc2d133f3ea
Signed-off-by: Manojkiran Eda <manojkiran.eda@gmail.com>

Currently, the ReleaseDateTime data is expected to be in a
"%d/%m/%Y %H:%M:%S" format.

However, this is ambiguous (ie., is locale-specific).

This change adds supoort to allow an ISO 8601 format of:

 "%Y-%m-%dT%H:%M:%S"

and a relaxed version with the 'T' separator replaced with a space:

 "%Y-%m-%d %H:%M:%S"

Update the readme and example json to switch to the preferred format.

Signed-off-by: Jeremy Kerr <jk@codeconstruct.com.au>
Change-Id: Ica3ae824979df6b666b8abb619ceaedffd90b374

Currently, all of the datetime parse is wrapped in a `catch KeyError`
block. However, the only part of this that would raise a KeyError is
where we extract the PackageReleaseDateTime from the metadata dict.

Instead, just do the key lookup separately, and parse outside of the
catch block.

Then, since we're handling a default anyway, just use .get() and then we
don't need to catch the KeyError.

Signed-off-by: Jeremy Kerr <jk@codeconstruct.com.au>
Change-Id: Iae6b92383af4fe49770e77005fcb575faa125703

pldm_entity_assocation_pdr_add() is deprecated in libpldm as it used
assert() for handling error cases.
pldm_entity_association_pdr_add_check() instead returns a value
indicating success or failure.

We throw a std::runtime_error for now which at least unwinds the stack,
unlike assert(). With this approach the problem is reduced to needing to
refactor internal pldmd APIs, rather than despairing at libpldm aborting
the process out of existence.

Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Change-Id: I08ebafc94cd7a392ec8f18b6c5feb75ba72cd028

To give more clarity, this commit adds an elaborate comment in
meson_options explaining why 5 seconds DBUS timeout value was
chosen for PLDM.

Change-Id: Icbfcb2c0521810ad1d31835ff8ccda67abd0e858
Signed-off-by: vkaverap@in.ibm.com <vkaverap@in.ibm.com>

This commit adds a new option "-n" to getPDR in the pldmtool using
which we can retrieve the PDRs based on the terminus ID.

Tested:
 ./pldmtool platform getpdr -h
get platform descriptor records
Usage: ./pldmtool platform GetPDR [OPTIONS]

Options:
  -h,--help                   Print this help message and exit
  -m,--mctp_eid UINT          MCTP endpoint ID
  -v,--verbose
[Option Group: Required Option]
  Retrieve individual PDR, all PDRs, PDRs of a requested type or retrieve all PDRs of the requested terminusID
  [Exactly 1 of the following options is required]
  Options:
    -d,--data UINT              retrieve individual PDRs from a PDR Repository
                                eg: The recordHandle value for the PDR to be retrieved and 0 means get first PDR in the repository.
    -t,--type TEXT              retrieve all PDRs of the requested type
                                supported types:
                                [terminusLocator, stateSensor, numericEffecter, stateEffecter, EntityAssociation, fruRecord, ... ]
    -i,--terminusID UINT          retrieve all PDRs of the requested terminusID
                                supported IDs:
                                 [1, 2, 208...]
    -a,--all                    retrieve all PDRs from a PDR repository

Example 1:
./pldmtool platform getpdr -i 0
[]

Example 2:
./pldmtool platform getpdr -i 2
Copied the output here:
https://gist.github.com/Pavithrab7/4218104bc7e58951ff91fe5b49e562b0



Change-Id: I8b78818abbbbb3678e180d2bd1f38017454fd0d3
Signed-off-by: Pavithra Barithaya <pavithra.b@ibm.com>

- Previously user had to do this configuration of selecting load
source, alternate load source and DASD load source manually,
and this was tedious task. These are network adapters connected
to PCI slots whose part location information is available in redfish.
- Adding bios attribute to handle selection of PCI slot location codes
for these network adapters to be selected from the GUI webpage, GUI
would then populate these BIOS attributes through redfish and this
selection will be communicated to Host via these PLDM BIOS attributes.

Change-Id: I7fa7cefe3323f4b21fb0335d0292ee136424fbe9
Signed-off-by: Kamalkumar Patel <kamalkumar.patel@ibm.com>

There is a lot of cruft left in the .gitignore from when this
repository was built using autotools and when it included libpldm,
so simplify it.

Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: Ice4f5fea1d0e19bea71643a56df06df547763642

Since enabling C++23, we can eliminate the need for `function2` and
instead leverage `std::move_only_function`.

Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: Iafc70c918e38b429e586bf18f71a716bdede43f8

The meson_options.txt looks to be very cumbersome and is difficult to
read & maintain due to in-consistent white spacing, and the lines are
longer than 80 characters.

This commit would fix those by migrating the syntax to new lines for
each option key, also warps the comments and the description to
80 characters & uses multilines for descriptions that are larger than 80
characters.Doing this should make this file more maintainable in future.

Change-Id: Ic8421cc21b8809c659be8e63b02b6877542a47fd
Signed-off-by: Manojkiran Eda <manojkiran.eda@gmail.com>

pldm_bios_table_attr_value_entry_encode_integer() is deprecated in
libpldm as it uses assert() to sanitize its parameters. Use
pldm_bios_table_attr_value_entry_encode_integer_check() instead, which
returns a value indicating success or failure.

Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Change-Id: Ic308d24a83fe3371a7ef074cb1d20092a06a1291

The error clang-diagnostic-overloaded-virtual is generated
when a function  hides the overloaded virtual function.

Change-Id: I87fdabb4ea9efd0c69d1b030127a6a9782cd3b23
Signed-off-by: Pavithra Barithaya <pavithra.b@ibm.com>

This error is generated when expression with side effects is evaluated
despite being used as an operand to 'typeid'.

Change-Id: I72daa47f1c5b05b466e1635fe6d547db4b1da34b
Signed-off-by: Pavithra Barithaya <pavithra.b@ibm.com>

This clang-error is generated when there is a constant unused
variable.

Change-Id: I4c101d3d58d468b728768fd63c1fd842e8c09ac4
Signed-off-by: Pavithra Barithaya <pavithra.b@ibm.com>

This clang-error is generated when a function overrides a member
function but is not marked 'override'.

Change-Id: I664941c2dff5be93dec1571cc974fae2168925f2
Signed-off-by: Pavithra Barithaya <pavithra.b@ibm.com>

The clang-diagnostic-defaulted-function-deleted error is generated when
explicitly defaulted move assignment operator is implicitly deleted.
This commit fixes those errors.

Change-Id: I7a407dd3411b19a73a13b1c1e4805b89fc2ec51a
Signed-off-by: Pavithra Barithaya <pavithra.b@ibm.com>

Meson 1.1.1 and GCC-13 both support C++23 and a sufficient portion of
the standard has been implemented.  Upgrade the build to leverage it.

Change-Id: I04732a67ef18c711ca80e53f0bf4828218f2b682
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>

pldm_pdr_add_fru_record_set() is deprecated in libpldm as it used
assert() to communicate failure to the caller.
pldm_pdr_add_fru_record_set_check() instead returns a value indicating
success or failure.

Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Change-Id: I88ed90a03b15a949641788906495c9507a5a196b

pldm_pdr_add() is deprecated in libpldm as it uses assert() to
communicate to the caller that an error occurred. pldm_pdr_add_check()
instead returns a value that indicates success or failure.

Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Change-Id: Icca92fd742f07fb935202a9d9e62f5648916b12b

pldm_entity_association_pdr_add_from_node() is deprecated in libpldm as
it relied on assert() to communicate failures from the implementation.
pldm_entity_association_pdr_add_from_node_check() instead returns a
value indicating success or failure.

Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Change-Id: If5ef1f372f0e835e8d97affcaab7a9abc2625d97

get_fru_record_by_option() is deprecated in libpldm as it relied on
assert() to communicate failures from the implementation.
get_fru_record_by_option_check() instead returns a value indicating
whether or not it succeeded.

Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Change-Id: I621bdcc713e37bb12ac40825edb399b151123427

As with handling pldm_pdr_init() failures, throw a std::runtime_error()
for pldm_entity_association_tree_init() failures. Throwing is an
improvement thanks to unwinding, by comparison to the immediate
termination invoked by assert().

Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Change-Id: I3d5d6bed5afa76503f16ec9d1842cf4fd09e9ae7

Migrate to pldm_bios_table_append_pad_checksum_check() which does not
sanitize its behaviour with assert().

Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Change-Id: Ide288bb40e5acabd2bf85870849bf176ddcbc621

This commit is to support a new VPD record called PSPD that contains
few keywords called #D and VM. Since these keywords are typically
larger in size (4K bytes) than what is managed by PLDM, we are adding
a file into which these keywords will be extracted. This keyword file
is transferred to the HB PLDM layer with a readFileByType request
from HB.

This also includes transfer of PSPD record data via PLDM with
offset included.

Tested by using pldmtool to send read request command to PLDM

Change-Id: Iba814a802cbbd2c2d641ee5a4c81919c3b8bdf20
Signed-off-by: Varsha Kaverappa <vkaverap@in.ibm.com>

Historically, libpldm's pldm_pdr_init() has internall used assert() to
enforce that the returned value is not NULL. Efforts are underway to
remove this reliance on assert() in libpldm to ensure consumers aren't
ungracefully terminated.

This change chooses to return an appropriate error to the
caller of pldmd's internal APIs where possible. However, in a number of
circumstances there was either not a sensible value that could be
returned, or it was not possible to return a value without further
refactoring. In these circumstances we choose to throw
`std::runtime_error`, which will at least do what it can to unwind.
While it may feel like it falls short of proper error handling, the
reality is the calling code is already structured such that it pretends
allocation failures never occur. The switch to throwing a runtime error
makes a strict improvement over the immediate termination implemented by
assert().

Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Change-Id: Ibb6bc3abf62e32f05ac079453853873298db6c4f

It's possible that the heap allocation invoked by
pldm_bios_table_iter_create() fails. The existing behaviour is to
assert() in the implementation that it hadn't failed, but that will be
changed to instead have the API return NULL.

Check for the NULL condition with the understanding that it may occur in
the future, and if observed, throw an exception to uphold the existing
assumptions in the code.

Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Change-Id: I20016e8db4c814017306cc68a536a9c232c74e0e

libpldm has deprecated the affected APIs. Prefer using their `*_check()`
variants as these return an error code rather than prevent misbehaviour
with assert().

1. pldm_bios_table_attr_entry_integer_encode()
2. pldm_bios_table_attr_value_entry_encode_enum()
3. pldm_bios_table_attr_value_entry_encode_string()
4. pldm_bios_table_attr_value_entry_encode_integer()

Tested: Successfully booted a p10bmc host, dumped each of the BIOS
        tables with pldmtool without issue

Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Change-Id: I823b24884e518f8c53fd153b3bc30569595ecd86

Change the interface xyz.openbmc_project.MCTP.Control to
xyz.openbmc_project.MCTP to fit the interface exposed by mctpd.service.

Also fixed the incorrect data type from size_t to mctp_eid_t.

Change-Id: Ic0267cef743615c1c0b253638d6c3fd390dae9ee
Signed-off-by: Delphine CC Chiu <Delphine_CC_Chiu@wiwynn.com>

Use of `dd(1)` results in back-to-back `read(2)` and `write(2)` syscalls;
the performance of `dd(1)` is O(n) with respect to the file size. Poor
choice of the `bs=` parameter to `dd(1)` can significantly amplify the
amount of syscalls required to achieve the task.

By contrast, `truncate(1)` generally invokes a single syscall, either
`truncate(2)` or `ftruncate(2)`. As a consequence, `truncate(1)` is O(1)
with respect to file size.

Switch to `truncate(1)` to improve runtime efficiency of the scripts.

Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Change-Id: I69dd925f36e733fc2a1ee71cb8354ee04ee8871b

The pldm_bios_table_*_check() APIs wrap around unchecked equivalents
that only sanitize their parameters using assert(). The checks included
in the pldm_bios_table_*_check() APIs test the same conditions that
would trigger the assert()s in the unchecked APIs but instead return an
error code on failure.

Use of the unchecked APIs has the potential to blow up the pldmd
process, but also requires that libpldm continue to be built with
`-UNDEBUG` to avoid undefined behaviour (by aborting the process). This
impacts the performance of the library in addition to be a bit of a UX
disaster.

The unchecked APIs are deprecated in libpldm as a step improve its
safety and code generation:

https://gerrit.openbmc.org/c/openbmc/libpldm/+/64179



Tested: Booted the host on a p10bmc system with the patch applied.

Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Change-Id: I9677a7fa5ca59d04865963b367f3bb55c8676cfb

This removes any ambiguity about which config.h is being included.
Subprojects such as libpldm and phosphor-logging also generate config.h
files, and these are exposed in the include path.

Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Change-Id: I3e0ccd6339f088380ffa29d41167b07aefc7dd16

Dbus timeout value for every dbus call in PLDM is set to 5 seconds
instead of default 25 seconds so that all dbus calls that expect a
reply exists after 5 seconds if there is no response from the client.
While BMC is still waiting for a response on any dbus calls it is
possible that host can timeout and return an error.
Another reason is that memory address from where BMC would read data
sent by host may get deallocated after 20 seconds and it is possible
for this to happen even before BMC has read the data. This has lead
to DMA issues.

Tested by adding a sleep for 10 seconds inside main function of
phospher-host-postd and executed get dbus property on snoopd
interface from pldm. PLDM dbus call timed out after 5 seconds.

Change-Id: I1aa1b4ad7fceff2224a3ccc3bcf6abd5705963bc
Signed-off-by: Varsha Kaverappa <vkaverap@in.ibm.com>

Traditional IBM systems since 1988 used to ship with A_Mode
as the default partition mode for iBMi. Changing the default
value from D_Mode to A_Mode to match the behaviour of our systems
and align them with the earlier behaviour.

Change-Id: I4fed505de3cf5628e5ad668870b6fd1ac00dc588
Signed-off-by: Manojkiran Eda <manojkiran.eda@gmail.com>

The existing code will allow ::<ipv4_address> without
any error. However, if the address is expanded like the
below, there will be "length out of range" error:
0000:0000:0000:0000:0000:ffff:xxx.xxx.xxx.xxx

This commit changes the maximum string length of
IPv6 address from 39 to 45, to support IPv4-mapped
IPv6 address.

Change-Id: I8c6e90e5a0ddbf38e7826e5918408b5d29811bfd
Signed-off-by: Asmitha Karunanithi <asmitk01@in.ibm.com>

This commit adds new UTIL record keywords RT, F5 and F6 under the
motherboard. F5 and F6 keywords are used to fill up the desired
host splash screen name displayed at the host console. The RT
keyword is used to hold the record name.

Tested: Verified that the name filled up through the F5 and F6
keywords are displayed at the host splash screen

Change-Id: I26f1c03db3e9f6a5b8cdc435d9e80d7c92563e16
Signed-off-by: Jayashankar Padath <jayashankar.padath@in.ibm.com>

With pldm commands like getPDR, pldmtool opens new socket for each PDR.
In case we have hundred number of PDRs, pldmtool open hundred number of
socket connection to server. It causes mctp-demux (role as server) reach
limit number of clients can handle.

So pldmtool should close socket connection to server after completely
using it.

Signed-off-by: Dung Cao <dung@os.amperecomputing.com>
Change-Id: I66fb00cd2d51f131c8901a8964552bcb2fc1de13

The host firmware (at least for IBM) never implemented this boot
progress when starting the OS. Instead, another host firmware team is
utilizing it to indicate an earlier part of the boot.

Therefore this boot progress code should not be treated as if the host
is running.

Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: I453ec41af804db880a99b62105b510464b147efa

In the pldmtool raw command, instanceId is inputted in raw data. It
does  not use the retrievedi one from the instanceId database by
exec() in pldm_cmd_helper. This causes the failure when the response
handler frees this provided instanceId because that instanceID is not
in libpldm database. To fix that, the inputted instanceId will be
ignored and the retrieved instanceId from libpldm will be used.

Tested:
1. No "Invalid instance ID" in the journal log when running "pldmtool
raw" commands.

Signed-off-by: Thu Nguyen <thu@os.amperecomputing.com>
Change-Id: I310f699e2d87ba2fa16b69350fdd7fa69d9598f9

The CI for new patch set report errors in softoff code.

'''
../softoff/softoff.cpp: In member function 'int pldm::SoftPowerOff::getSensorInfo()':
../softoff/softoff.cpp:263:14: error: 'pdr' may be used uninitialized [-Werror=maybe-uninitialized]
  263 |         auto possibleStatesStart = pdr->possible_states;
      |              ^~~~~~~~~~~~~~~~~~~
../softoff/softoff.cpp:249:32: note: 'pdr' was declared here
  249 |         pldm_state_sensor_pdr* pdr;
cc1plus: all warnings being treated as errors
'''

Update code to fix that.

Signed-off-by: ThuBaNguyen <thu@os.amperecomputing.com>
Change-Id: Ib9df9f79978941ebad62164866cc688412a08922

and remove unnecessary `struct` from class variable;

Change-Id: Id7e76bd8f6123cc47d27cfc232a2a9186f2cad06
Signed-off-by: Rashmica Gupta <rashmica@linux.ibm.com>