Conceptually we were trying to stuff subtly different actions into one vfunc:
* Read firmware from the device to update the verification checksums
* Read a firmware blob from the device for debugging
For the first action we might want to mask out the sections of the flash with
serial numbers (so the verification hashes match the ones published on the LVFS)
and for the second we want just a raw ROM file from the hardware with no
pre-processing that we can compare against an external SPI dumper.
Split out ->dump_firmware to get the raw blob, and allow plugins to also
implement ->read_firmware() if they have to mask out specific offsets or remove
specific images from the FuFirmware container.
In the common case when masking is not required, fall back to using a 'binary'
FuFirmware automatically to make most plugins simpler.
At the moment there are commands to convert one file format to another, but not
to 'merge' or alter them. Some firmware files are containers which can store
multiple images, each with optional id, idx and addresses.
This would allow us to, for instance, create a DfuSe file with two different
raw files that are flashed to different addresses on the SPI flash. It would
also allow us to create very small complicated container formats for fuzzing.
This can be used by writing a `firmware.builder.xml` file like:
<?xml version="1.0" encoding="UTF-8"?>
<firmware gtype="FuBcm57xxFirmware">
<version>1.2.3</version>
<image>
<version>4.5.6</version>
<id>header</id>
<idx>456</idx>
<addr>0x456</addr>
<filename>header.bin</filename>
</image>
<image>
<version>7.8.9</version>
<id>payload</id>
<idx>789</idx>
<addr>0x789</addr>
<data>aGVsbG8=</data>
</image>
</firmware>
...and then using something like:
# fwupdtool firmware-convert firmware.builder.xml firmware.dfu builder dfu
This is only applicable for kernel 5.9 or kernels that backported
the authenticate on disconnect patches.
For installation time this isn't very important since no device will
restart. This also isn't relevant for the authenticate on disconnect
scenario.
However the manual activation scenario, it's important to activate the
WD19TB device first, followed by Thunderbolt.
I2C doesn't have any specification for what is a probe and what is a more
destructive action. Sending tx_buf out on the i2c bus to a generic address
might not be safe in all cases.
To prevent this, use a HWID to check the machine DMI value during device
creation before ->probe() or ->setup() is called on the device.
If the device is not authorized, it may cause a composite update that it's part
of to not behave properly.
If device is authorized at runtime, add updatable flag at runtime as well
See #2374 for more details
This ensures we perform the updates in this order:
* cxaudio
* vli
* thunderbolt
As any other order causes enumeration failures.
Fixes some of https://github.com/fwupd/fwupd/issues/2377
Error will show up if (priv->physical_id == NULL) on
fu_device_ensure_id(). This method is called after probe but before
setup on fu_device_open.
Call fu_device_set_physical_id() on fu_thunderbolt_device_probe()
for the retimer case.
Fixes https://github.com/fwupd/fwupd/issues/2371
Change-Id: I0e462fff5e8abf6073318f6424b6736afc8259b8
This object derives from FuDevice rather than FuVliDevice and so does not
inherit the NO_GUID_MATCHING flag like other devices.
Fixes https://github.com/fwupd/fwupd/issues/2364
It seems the MEI parameters are not set for platforms such as Apollo Lake.
The only bit set was FPF_SOC_LOCK, with the others all cleared to zero.
Fixes https://github.com/fwupd/fwupd/issues/2335 for a peculiar definition of 'fix'.
As described in DFU protocol (7. Manifestation Phase), after the
firmware reprogramming is done, if bitWillDetach = 1, it doesn't
require the host to issue a USB bus reset, but the device can
generate a detach-attach sequence itself to go back to normal.
Add a quirk flag "no-bus-reset-attach" to skip the bus reset in
dfu_device_attach(), and increase the "RemoveDelay" as well.
Print the sysfs path for devices deriving from FuUdevDevice, which also allows
us to use FU_UDEV_DEVICE_DEBUG without monkey-patching the plugins that also
define a device_class->to_string() vfunc.
For some Poly USB Cameras, it takes a longer time than the
default (FU_DEVICE_REMOVE_DELAY_RE_ENUMERATE) for being detached
to DFU or attached to normal mode. Need to specify the timeout in
"RemoveDelay" quirk key.
Also replace the hard-coded timeout with fu_device_get_remove_delay()
in dfu-tool.c.
Some devices may accumulate the firmware image and perform the
entire reprogramming operation at one time. In this case, the
device enters dfuMANIFEST-SYNC or dfuMANIFEST state after
dfuDNLOAD-IDLE.
The fwupd shall be able to poll the status from the device via
DFU_GETSTATUS until the device completes the reprogramming or
reports an error.
For details, please refer to Section 7. Manifestation Phase and
A.1 Interface State Transition Diagram in the USB DFU protocol.
https://www.usb.org/sites/default/files/DFU_1.1.pdf
For not affecting the other DFU capable devices, introduce a quirk
"manifest-poll" to limit the logic.
The nr_chunks is defined as an unsigned short, the max value is
65536. Assume the transfer_size reported by device is 4096, the
maximum size of DFU firmware supported is 65536 * 4096 = 256MB.
To support larger DFU firmware, we can change the guint16 to
guint32.
This implements the following simple state machine:
1. Reboot to RO, Update RW first, set ANOTHER_WRITE_REQUIRED
2. Second time around, don't reboot into RO. Stay in RW.
3. Update RO.
4. Done.
Only add instance ID if it actually probes properly.
Otherwise this makes an invalid assumption that the device is a WD19
EC just because it had the correct hub in front.
Instead check the first time it's opened that the correct device
is identified (`EXPECTED_DOCK_TYPE`)
This is much more efficient than parsing hundreds of lines of /proc/cpuinfo
and also causes hundreds of thousands less allocations at startup. For systems
with dozens of virtual CPUs the deduplication of device objects was increasing
start up time considerably.
Use the msr plugin to read the microcode version as this is not obtained using
CPUID, as it is instead being provided in an MSR.
To do this mount all ESP partitions and check all the binaries there to see if
they match any entries in the new dbx. If we applied the update when a hash
matched, we would unintentially 'brick' the users machine, as the grub and shim
binaries *have* to be updated first.
This functionality does reimplement the PE hashing functionality found in
sbsigntools and pesign. This was done for 4 main reasons:
* There were some memory safety issues found when fuzzing random binaries
* Executing the tools hundreds of times was a lot of overhead
* Operating from a blob of immutable mmap'd memory is much faster
* We only need a very small amount of functionality from both tools
The I²C proxy specification is not always shared with all other devices as
I originally hoped, instead there are other legacy devices that use different
sets of I²C commands.
Un-share various bits of code to allow for additional I²C devices to be added.
No logic changes.
Write the firmware to the usb device, first finding which section to
write to, then breaking into blocks (based on maximum pdu size),
and then into chunks, which are transferred to the device using bulk transfers.
The register specifications have been taken as a superset of the coreboot
documentation as different flags were documented in more detail on various
different platforms.
Having this new data allows us to add future tests and make the current tests
much easier to understand.
This makes perfect sense, because the 'initiator' starts the transaction and
the 'target' is the addressee of the transaction. Even the I²C spec defines the
'master' as 'initiating' the transaction.
This is the same nomenclature now used by the Glasgow project too.
This was an overloaded use of UpdateMessage that didn't make sense.
It doesn't affect the functionality of updating, just the security.
Hints about why the TPM PCR0 reconstruction failed should go
to the wiki page not the device.
We'll instead check this when the user tries to run an update. This
allows them to sign a bootloader after the daemon starts (or remove
a signed bootloader after starting)
Fixes: #2219
Trying to explain why ICL thunderbolt isn't updatable doesn't help
people. It just causes fwupdmgr and fwupdtool to show the device
front and center with a confusing message.
Instead don't populate the message and by the default device filter
it will be hidden.
See #2212 for background.
Reading the sysfs file seemed to have also eaten the `\n` as mentioned
on a bug.
```
├DW5821e Snapdragon X20 LTE:
│ Device ID: fa707b9af86ff44bc17316b6c3e5ea82aab3ce86
│ Summary: Mobile broadband device
│ Current version: T77W968.F1.0.0.4.2.GC.010
│ Vendor: Dell Inc. (USB:0x413c
│ )
│ GUIDs: 64da2d58-8d1b-5e5b-b793-f88ba5a25a8f
│ 761d6124-0002-5185-b767-9adf67bf1a5e
│ 795e079d-093b-5503-aa59-35b832480e95
│ Device Flags: • Updatable
```
This allows delaying the activation of Thunderbolt firmware until
shutdown/reboot or when the dock is unplugged.
This functionality requires features in the kernel:
https://lore.kernel.org/linux-usb/20200622143035.25327-1-mario.limonciello@dell.com/T/#t
Matrix of cases to support:
* Distro Old Linux kernel (doesn't support authenticate on disconnect)
- WD19TB: Should have `skips-restart` flag set
No flush or activate features called in `thunderbolt` plugin.
`dell_dock` plugin will activate at end of composite update
- All other devices: Shouldn't have flags set
Should authenticate in Thunderbolt plugin.
`1 > nvm_authenticate`
* Distro New Linux kernel (supports authenticate on disconnect)
- WD19TB: Should have `usable-during-update` flag set but not `skips-restart`
Should flush image to SPI in `thunderbolt` plugin
`2 > nvm_authenticate_on_disconnect`
Should configure TBT device for authenticate on disconnect
`1 > nvm_authenticate_on_disconnect`
`dell_dock` plugin will configure dock for authenticate on disconnect
- All other devices: Shouldn't have flags set
Should authenticate in `thunderbolt` plugin.
`1 > nvm_authenticate`
* ChromeOS (supports authenticate on disconnect)
- `thunerbolt.conf` will have `DelayedActivation=true`.
- WD19TB: Should have `usable-during-update` flag set but not `skips-restart`
Should flush image to SPI in `thunderbolt` plugin
`2 > nvm_authenticate_on_disconnect`
Should configure device for authenticate on disconnect
`1 > nvm_authenticate_on_disconnect`
`dell_dock` plugin will configure dock for authenticate on disconnect
- All other devices: Should have both `usable-during-update` and `skips-restart` set
Should flush image to SPI in `thunderbolt` plugin
`2 > nvm_authenticate`
Will activate upon logout/shutdown/reboot
`1 > nvm_authenticate`
This plugin is only enabled when coreboot isn't detected.
It intentionally does not check for EFI to be disabled at startup
since it can also notify the user that UEFI capsule updates are
disabled on the system even if running in UEFI mode.
Unfortunately module type has more than I previously realized.
The meanings that previously were applied fortunately worked for
the most important case (130-180W TBT) but didn't for single C, dual
C or small power (45W) cases.
Since composite_prepare was trying to read and interpret these, it
causes failures when these other ones are encountered.
I reproduced this on a 130W adapter plugged into a single C (type 0x4).
This meant the update wouldn't install since NULL was returned for the
type.
In case a new module ID is added later, also return an "unknown" for
the metadata.
Add two new vfuncs that can be used to collect report metadata from devices
both before and after the update has run. This means we can remove the hacks
where we set add 'global' metadata entries and just hope that there is only one
device from the same plugin that is updated.
This also allows us to collect debugging metadata from devices after an offline
update has been run.
As both hub devices share a FuVliUsbhubDeviceClass instance we cannot 'hijack'
the vfuncs depending on object type. This allows the downstream hub to proxy to
the upstream hub where a GPIOB reset can be performed.
We can have multiple FuVliPdDevice objects registered with the daemon, but they
will all share the FuVliPdDeviceClass instance. If one device requries a
silicon workaround, do not 'hijack' the vfunc for all devices of this type.
This means we do the right thing when updating both the one that requires the
workaround, and the 'normal' one.
Provide the following metadata:
"Dirty firmware" bit
Protocol version
Header type
Maximum PDU Size
Flash protection status
Raw version string
Key Version
Minimum rollback
This should do the bulk transfers using protocol 6.
The output here is now equivalent to the output of
usb_updater2 -d 18d1:501a -f
fwupdtool --plugin-whitelist cros_ec get-devices --verbose
Servo Micro
DeviceId: 84d0e3f2a0f8b2328f7995767b23ebb40494723f
Guid: 8e2f7625-a164-55d7-8f09-f193c8ec33f1 <- USB\VID_18D1&PID_501A&REV_0100
Guid: 13564257-c649-586d-b4e4-4f048d480f36 <- USB\VID_18D1&PID_501A
Serial: CMO653-00166-040491U00771
Summary: Servo Micro (aka "uServo") Debug Board
Plugin: cros_ec_usb
Flags: registered
Vendor: Google Inc.
VendorId: USB:0x18D1
Version: servo_micro_v2.4.17-df61092c3
VersionFormat: plain
Created: 2020-05-20
PhysicalId: usb:03:00:02
Although they normally work, some failures have been reported in the field
related to the MST hub not responding in the MST plugin.
When these failures have occurred the dell_dock plugin also fails to
enumerate.
So rather than allow some people who don't have dell_dock compiled to
update their MST hub using synaptics_mst, perform ALL updates for
mst hub via dell_dock.
```
18:06:24:0324 FuPluginSynapticsMST no device found on drm_dp_aux1: VMM5331 inside Dell dock is only supported by dell_dock
```
See also: https://github.com/fwupd/fwupd/issues/2119
Reported-by: Anton Farygin <rider@altlinux.org>
Based on a patch by Gleb Fotengauer-Malinovskiy <glebfm@altlinux.org>
Fixes cross compilation for arm with clang which provides cpuid.h
but helpfully has this error:
```
#if !(__x86_64__ || __i386__)
#error this header is for x86 only
#endif
```
Fixes: #2131
To do this, rely on the AppStream ID to map to a translated string (providing a
fallback for clients that do not care) and switch the free-form result string
into a set of enumerated values that can be translated.
This fixes some of the problems where some things have to be enabled to "pass"
and other attributes have to be some other state. For cases where we want the
user to "do" something, provide a URL to a wiki page that we update out-of-band
of fwupd releases.
This only checks that it was available from the CPU.
To be complete an additional check should be made to show that it
was actually enabled from the firmware.
This will require a kernel modification though because MSR access
will be forbidden from userland while in kernel lockdown.
The kernel patches are a log way from being upstreamed, so disable this until
there is even a chance the user might be running it.
This removes the obsoletes line from *every* system running 'fwupdmgr security'.
We can read this from userspace even when SB is turned on and with the kernel
locked down. The kernel securityfs patches are still in-progress, but will take
significant time to get upstream.
The kernel patches are needed when the PCI device is hidden from userspace.
On a system that is not at all locked down running an old kernel several
of the items are a bit confusing.
```
Runtime Suffix -!
✔ fwupd plugins: OK
✔ Linux Kernel: OK
✘ Linux Kernel: Could not open file
✘ Linux Swap: Not encrypted
```
The HSI specification is currently incomplete and in active development.
Sample output for my Lenovo P50 Laptop:
Host Security ID: HSI:2+UA!
HSI-1
✔ UEFI dbx: OK
✔ TPM: v2.0
✔ SPI: Write disabled
✔ SPI: Lock enabled
✔ SPI: SMM required
✔ UEFI Secure Boot: Enabled
HSI-2
✔ TPM Reconstruction: Matched PCR0 reading
HSI-3
✘ Linux Kernel S3 Sleep: Deep sleep available
HSI-4
✘ Intel CET: Unavailable
Runtime Suffix -U
✔ Firmware Updates: Newest release is 8 months old
Runtime Suffix -A
✔ Firmware Attestation: OK
Runtime Suffix -!
✔ fwupd plugins: OK
✔ Linux Kernel: OK
✔ Linux Kernel: Locked down
✘ Linux Swap: Not encrypted
This exports FuSecurityAttrs into libfwupdplugin so that we can pass the plugins
this object rather than a 'bare' GPtrArray. This greatly simplifies the object
ownership, and also allows us to check the object type before adding.
In the future we could also check for duplicate appstream IDs or missing
properties at insertion time.
This change also changes the fu_plugin_add_security_attrs() to not return an
error. This forces the plugin to handle the error, storing the failure in the
attribute itself.
Only the plugin know if a missing file it needs to read indicates a runtime
problem or a simple failure to obtain a specific HSI level.
MATEKF722SE has unconvetional behavior for dfu protocol, where the sector size
isn't specified and sector type is shiffted left by 1. This happens only for
one sector.
Sector parsing from MATEKF722SE:
* `016Kg`
* `64Kg`
* `128Kg`
* `048 e`
* `528e`
* `004 e`
