iPXE will currently fail all SNP interface methods with EFI_NOT_READY
while the network devices are claimed for use by iPXE's own network
stack.
As of commit c70b3e0 ("[efi] Always enable recursion when calling
ConnectController()"), this exposes latent UEFI firmware bugs on some
systems at the point of calling ExitBootServices().
With recursion enabled, the MnpDxe driver will immediately attempt to
consume the SNP protocol instance provided by iPXE. Since the network
devices are claimed by iPXE at this point, the calls by MnpDxe to
Start() and Initialize() will both fail with EFI_NOT_READY.
This unfortunately triggers a broken error-handling code path in the
Ip6Dxe driver. Specifically: Ip6DriverBindingStart() will call
Ip6CreateService(), which will call Ip6ServiceConfigMnp(), which will
return an error. The subsequent error handling code path in
Ip6CreateService() simply calls Ip6CleanService(). The code in
Ip6CleanService() will attempt to leave the all-nodes multicast group,
which will fail since the group was never joined. This will result in
Ip6CleanService() returning an error and omitting most of the required
clean-up operations. In particular, the MNP protocol instance will
remain opened with BY_DRIVER attributes even though the Ip6Dxe driver
start method has failed.
When ExitBootServices() is eventually called, iPXE will attempt to
uninstall the SNP protocol instance. This results in the UEFI core
calling Ip6DriverBindingStop(), which will fail since there is no
EFI_IP6_SERVICE_BINDING_PROTOCOL instance installed on the handle.
A failure during a call to UninstallMultipleProtocolInterfaces() will
result in the UEFI core attempting to reinstall any successfully
uninstalled protocols. This is an intrinsically unsafe operation, and
represents a fundamental design flaw in UEFI. Failure code paths
cannot be required to themselves handle failures, since there is no
well-defined correct outcome of such a situation.
With a current build of OVMF, this results in some unexpected debug
messages occurring at the time that the loaded operating system calls
ExitBootServices(). With the UEFI firmware in Hyper-V, the result is
an immediate reboot.
Work around these UEFI design and implementation flaws by allowing the
calls to our EFI_SIMPLE_NETWORK_PROTOCOL instance's Start() and
Initialize() methods to return success even when the network devices
are claimed for exclusive use by iPXE. This is sufficient to allow
MnpDxe to believe that it has successfully initialised the device, and
thereby avoids the problematic failure code paths in Ip6Dxe.
Debugged-by: Aaron Heusser <aaron_heusser@hotmail.com>
Debugged-by: Pico Mitchell <pico@randomapplications.com>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
For the physical function driver, the transmit queue needs to be
configured to be associated with the relevant physical function
number. This is currently obtained from the bus:dev.fn address of the
underlying PCI device.
In the case of a virtual machine using the physical function via PCI
passthrough, the PCI bus:dev.fn address within the virtual machine is
unrelated to the real physical function number. Such a function will
typically be presented to the virtual machine as a single-function
device. The function number extracted from the PCI bus:dev.fn address
will therefore always be zero.
Fix by reading from the Function Requester ID Information Register,
which always returns the real PCI bus:dev.fn address as used by the
physical host.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The datasheet is fairly incomprehensible in terms of identifying the
appropriate MAC address for use by the physical function driver.
Choose to read the MAC address from PRTPM_SAH and PRTPM_SAL, which at
least matches the MAC address as selected by the Linux i40e driver.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
iPXE will currently drop to TPL_APPLICATION whenever the current
system time is obtained via currticks(), since the system time
mechanism relies on a timer that can fire only when the TPL is below
TPL_CALLBACK.
This can cause unexpected behaviour if the system time is obtained in
the middle of an API call into iPXE by external code. For example,
MnpDxe sets up a 10ms periodic timer running at TPL_CALLBACK to poll
the underling EFI_SIMPLE_NETWORK_PROTOCOL device for received packets.
If the resulting poll within iPXE happens to hit a code path that
requires obtaining the current system time (e.g. due to reception of
an STP packet, which affects iPXE's blocked link timer), then iPXE
will end up temporarily dropping to TPL_APPLICATION. This can
potentially result in retriggering the MnpDxe periodic timer, causing
code to be unexpectedly re-entered.
Fix by recording the external TPL at any entry point into iPXE and
dropping only as far as this external TPL, rather than dropping
unconditionally to TPL_APPLICATION.
The side effect of this change is that iPXE's view of the current
system time will be frozen for the duration of any API calls made into
iPXE by external code at TPL_CALLBACK or above. Since any such
external code is already responsible for allowing execution at
TPL_APPLICATION to occur, then this should not cause a problem in
practice.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Physical addresses in debug messages are more meaningful from an
end-user perspective than potentially IOMMU-mapped I/O virtual
addresses, and have the advantage of being calculable without access
to the original DMA mapping entry (e.g. when displaying an address for
a single failed completion within a descriptor ring).
Signed-off-by: Michael Brown <mcb30@ipxe.org>
For a software UNDI, the addresses in PXE_CPB_TRANSMIT.FrameAddr and
PXE_CPB_RECEIVE.BufferAddr are host addresses, not bus addresses.
Remove the spurious (and no-op) use of virt_to_bus() and replace with
a cast via intptr_t.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The UEFI specification defines PXE_CPB_TRANSMIT.DataLen as excluding
the length of the media header. iPXE currently fills in DataLen as
the whole frame length (including the media header), along with
placing the media header length separately in MediaheaderLen. On some
UNDI implementations (observed using a VMware ESXi 7.0b virtual
machine), this causes transmitted packets to include 14 bytes of
trailing garbage.
Match the behaviour of the EDK2 SnpDxe driver, which fills in DataLen
as the whole frame length (including the media header) and leaves
MediaheaderLen as zero. This behaviour also violates the UEFI
specification, but is likely to work in practice since EDK2 is the
reference implementation.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
iPXE currently assumes that DMA-capable devices can directly address
physical memory using host addresses. This assumption fails when
using an IOMMU.
Define an internal DMA API with two implementations: a "flat"
implementation for use in legacy BIOS or other environments in which
flat physical addressing is guaranteed to be used and all allocated
physical addresses are guaranteed to be within a 32-bit address space,
and an "operations-based" implementation for use in UEFI or other
environments in which DMA mapping may require bus-specific handling.
The purpose of the fully inlined "flat" implementation is to allow the
trivial identity DMA mappings to be optimised out at build time,
thereby avoiding an increase in code size for legacy BIOS builds.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The malloc_dma() function allocates memory with specified physical
alignment, and is typically (though not exclusively) used to allocate
memory for DMA.
Rename to malloc_phys() to more closely match the functionality, and
to create name space for functions that specifically allocate and map
DMA-capable buffers.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Provide opened EFI PCI devices with access to the underlying
EFI_PCI_IO_PROTOCOL instance, in order to facilitate the future use of
the DMA mapping methods within the fast data path.
Do not require the use of this stored EFI_PCI_IO_PROTOCOL instance for
memory-mapped I/O (since the entire point of memory-mapped I/O as a
concept is to avoid this kind of unnecessary complexity) or for
slow-path PCI configuration space accesses (since these may be
required for access to PCI bus:dev.fn addresses that do not correspond
to a device bound via our driver binding protocol instance).
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The legacy transmit descriptor index is not reset by anything short of
a full device reset. This can cause the legacy transmit ring to stall
after closing and reopening the device, since the hardware and
software indices will be out of sync.
Fix by performing a reset after closing the interface. Do this only
if operating in legacy mode, since in C+ mode the reset is not
required and would undesirably clear additional state (such as the C+
command register itself).
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Some UEFI systems (observed with a Supermicro X11SPG-TF motherboard)
seem to fail to provide a valid ACPI address space descriptor for the
MMIO address space associated with a PCI root bridge.
If no valid descriptor can be found, fall back to assuming that the
MMIO address space is identity mapped, thereby matching the behaviour
prior to commit 27e886c ("[efi] Use address offset as reported by
EFI_PCI_ROOT_BRIDGE_IO_PROTOCOL").
Debugged-by: Tore Anderson <tore@fud.no>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Commit 87e39a9c9 ("[efi] Split efi_usb_path() out to a separate
function") unintentionally introduced an undefined symbol reference
from efi_path.o to usb_depth(), causing the USB subsystem to become a
dependency of all EFI builds.
Fix by converting usb_depth() to a static inline function.
Reported-by: Pico Mitchell <pico@randomapplications.com>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The UEFI specification allows uninstallation of a protocol interface
to fail. There is no sensible way for code to react to this, since
uninstallation is likely to be taking place on a code path that cannot
itself fail (e.g. a code path that is itself a failure path).
Where the protocol structure exists within a dynamically allocated
block of memory, this leads to possible use-after-free bugs. Work
around this unfortunate design choice by nullifying the protocol
(i.e. overwriting the method pointers with no-ops) and leaking the
memory containing the protocol structure.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Use the device path constructed via efi_describe() for the installed
EFI_BLOCK_IO_PROTOCOL device handle.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The UEFI specification provides a partial definition of an Infiniband
device path structure. Use this structure to construct what may be a
plausible path containing at least some of the information required to
identify an SRP target device.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The ACPI table contents are typically large and are likely to cause
any preceding error messages to scroll off-screen.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
There is no standard defined for AoE device paths in the UEFI
specification, and it seems unlikely that any standard will be adopted
in future.
Choose to construct an AoE device path using a concatenation of the
network device path and a SATA device path, treating the AoE major and
minor numbers as the HBA port number and port multiplier port number
respectively.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Provide efi_netdev_path() as a standalone function, to allow for reuse
when constructing child device paths.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Allow an interface operation to be declared as unused. This will
perform full type-checking and compilation of the implementing method,
without including any code in the resulting object (other than a NULL
entry in the interface operations table).
The intention is to provide a relatively clean way for interface
operation methods to be omitted in builds for which the operation is
not required (such as an operation to describe an object using an EFI
device path, which would not be required in a non-EFI build).
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Now that IPv6 is enabled by default for UEFI builds, it is important
that iPXE does not delay unnecessarily in the (still relatively
common) case of a network that lacks IPv6 routers.
Apply the timeout values used for neighbour discovery to the router
discovery process.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
IPv6 PXE was included in the UEFI specification over eight years ago,
specifically in version 2.3 (Errata D).
http://www.uefi.org/sites/default/files/resources/UEFI_Spec_2_3_D.pdf
When iPXE is being chainloaded from a UEFI firmware performing a PXE
boot in an IPv6 network, it is essential that iPXE supports IPv6 as
well.
I understand that the reason for NET_PROTO_IPV6 being disabled by
default (in src/config/general.h) is that it would cause certain
space-constrained build targets to become too large. However, this
should not be an issue for EFI builds.
It is also worth noting that RFC 6540 makes a clear recommendation
that IPv6 support should not be considered optional.
https://tools.ietf.org/html/rfc6540
Modified-by: Michael Brown <mcb30@ipxe.org>
Signed-off-by: Tore Anderson <tore@fud.no>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The LACP responder reuses the received I/O buffer to construct the
response LACP (or marker) packet. Any received padding will therefore
be unintentionally included within the response.
Truncate the received I/O buffer to the expected length (which is
already defined in a way to allow for future protocol expansion)
before reusing it to construct the response.
Reported-by: Tore Anderson <tore@fud.no>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Some external drivers (observed with the UEFI NII driver provided by
an HPE-branded Mellanox ConnectX-3 Pro) seem to cause LACP packets
transmitted by iPXE to be looped back as received packets. Since
iPXE's trivial LACP responder will send one response per received
packet, this results in an immediate LACP packet storm.
Detect looped back LACP packets (based on the received LACP actor MAC
address), and refuse to respond to such packets.
Reported-by: Tore Anderson <tore@fud.no>
Tested-by: Tore Anderson <tore@fud.no>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
When iPXE is downloading a file from an EFI_FILE_PROTOCOL instance
backed by an EFI_BLOCK_IO_PROTOCOL instance provided by the same iPXE
binary (e.g. via a hooked SAN device), then it is possible for step()
to be invoked as a result of the calls into the EFI_BLOCK_IO_PROTOCOL
methods. This can potentially result in efi_local_step() being run
prematurely, before the file has been opened and before the parent
interface has been attached.
Fix by deferring starting the download process until immediately prior
to returning from efi_local_open().
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Some UEFI BIOSes (observed with at least the Insyde UEFI BIOS on a
Microsoft Surface Go) provide a very broken version of the
UsbMassStorageDxe driver that is incapable of binding to the standard
EFI_USB_IO_PROTOCOL instances and instead relies on an undocumented
proprietary protocol (with GUID c965c76a-d71e-4e66-ab06-c6230d528425)
installed by the platform's custom version of UsbCoreDxe.
The upshot is that USB mass storage devices become inaccessible once
iPXE's native USB host controller drivers are loaded.
One possible workaround is to load a known working version of
UsbMassStorageDxe (e.g. from the EDK2 tree): this driver will
correctly bind to the standard EFI_USB_IO_PROTOCOL instances exposed
by iPXE. This workaround is ugly in practice, since it involves
embedding UsbMassStorageDxe.efi into the iPXE binary and including an
embedded script to perform the required "chain UsbMassStorageDxe.efi".
Provide a native USB mass storage driver for iPXE, allowing USB mass
storage devices to be exposed as iPXE SAN devices.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
iPXE will often have multiple drivers available for a USB device. For
example: some USB network devices will support both RNDIS and CDC-ECM,
and any device may be consumed by the fallback "usbio" driver under
UEFI in order to expose an EFI_USB_IO_PROTOCOL instance.
The driver scoring mechanism is used to select a device configuration
based on the availability of drivers for the interfaces exposed in
each configuration.
For the case of RNDIS versus CDC-ECM, this mechanism will always
produce the correct result since RNDIS and CDC-ECM will not exist
within the same configuration and so each configuration will receive a
score based on the relevant driver.
This guarantee does not hold for the "usbio" driver, which will match
against any device. It is a surprising coincidence that the "usbio"
driver seems to usually end up at the tail end of the USB drivers
list, thereby resulting in the expected behaviour.
Guarantee the expected behaviour by explicitly placing the "usbio"
driver at the end of the USB drivers list.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
For USB mass storage devices, we do not want to submit more bulk IN
packets than are required for the inbound data, since this will waste
memory.
Allow an upper limit to be specified on each refill attempt. The
endpoint will be refilled to the lower of this limit or the limit
specified by usb_refill_init().
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Closing and reopening a USB endpoint will clear any halt status
recorded by the host controller, but may leave the endpoint halted at
the device. This will cause the first packet submitted to the
reopened endpoint to be lost, before the automatic stall recovery
mechanism detects the halt and resets the endpoint.
This is relatively harmless for USB network or HID devices, since the
wire protocols will recover gracefully from dropped packets. Some
protocols (e.g. for USB mass storage devices) assume zero packet loss
and so would be adversely affected.
Fix by allowing any device endpoint halt status to be cleared on a
freshly opened endpoint.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
There appears to be no reason for avoiding recursion when calling
ConnectController(), and recursion provides the least surprising
behaviour.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Michael Brown
Tore Anderson