Use the tail register offset (which exists for all ring types) as the
ring identifier in all relevant debug messages.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
For the sake of completeness, ensure that all 32 bytes of the receive
queue context are programmed (including the unused final 8 bytes).
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Commit 8f3e648 ("[intelxl] Use one admin queue buffer per admin queue
descriptor") changed the API for intelxl_admin_command() such that the
caller now constructs the command directly within the next available
descriptor ring entry, rather than relying on intelxl_admin_command()
to copy the descriptor to and from the descriptor ring.
This introduced a regression in intelxl_admin_switch(), since the
second and subsequent iterations of the loop will not have constructed
a valid command in the new descriptor ring entry before calling
intelxl_admin_command().
Fix by constructing the command within the loop.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
When applying an ACPI-provided system-specific MAC address, apply it
to netdev->ll_addr rather than netdev->hw_addr. This allows iPXE
scripts to access the permanent MAC address via the ${netX/hwaddr}
setting (and thereby provides scripts with a mechanism to ascertain
that the NIC is using a MAC address other than its own permanent
hardware address).
Signed-off-by: Michael Brown <mcb30@ipxe.org>
As per the general pattern for initialisation functions in iPXE,
pci_init() saves code size by assuming that the caller has already
zeroed the underlying storage (e.g. as part of zeroing a larger
containing structure). There are several places within the code where
pci_init() is deliberately used to initialise a transient struct
pci_device without zeroing the entire structure, because the calling
code knows that only the PCI bus:dev.fn address is required to be
initialised (e.g. when reading from PCI configuration space).
Ensure that using pci_init() followed by pci_read_config() will fully
initialise the struct pci_device even if the caller did not previously
zero the underlying storage, since Coverity reports that there are
several places in the code that rely upon this.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Several keyboard layouts define ASCII characters as accessible only
via the AltGr modifier. Add support for this modifier to ensure that
all ASCII characters are accessible.
Experiments suggest that the BIOS console is likely to fail to
generate ASCII characters when the AltGr key is pressed. Work around
this limitation by accepting LShift+RShift (which will definitely
produce an ASCII character) as a synonym for AltGr.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Handle Ctrl and CapsLock key modifiers within key_remap(), to provide
consistent behaviour across different console types.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The key with scancode 86 appears in the position between left shift
and Z on a US keyboard, where it typically fails to exist entirely.
Most US keyboard maps define this nonexistent key as generating "\|",
with the notable exception of "loadkeys" which instead reports it as
generating "<>". Both of these mapping choices duplicate keys that
exist elsewhere in the map, which causes problems for our ASCII-based
remapping mechanism.
Work around these quirks by treating the key as generating "\|" with
the high bit set, and making it subject to remapping. Where the BIOS
generates "\|" as expected, this allows us to remap to the correct
ASCII value.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Some keyboard layouts (e.g. "fr") swap letter and punctuation keys.
Apply the logic for upper and lower case and for Ctrl-<key> only after
applying remapping, in order to handle these layouts correctly.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
On some systems (observed with the Thunderbolt ports on a ThinkPad X1
Extreme Gen3 and a ThinkPad P53), if the IOMMU is enabled then the
system firmware will install an ExitBootServices notification event
that disables bus mastering on the Thunderbolt xHCI controller and all
PCI bridges, and destroys any extant IOMMU mappings. This leaves the
xHCI controller unable to perform any DMA operations.
As described in commit 236299b ("[xhci] Avoid DMA during shutdown if
firmware has disabled bus mastering"), any subsequent DMA operation
attempted by the xHCI controller will end up completing after the
operating system kernel has reenabled bus mastering, resulting in a
DMA operation to an area of memory that the hardware is no longer
permitted to access and, on Windows with the Driver Verifier enabled,
a STOP 0xE6 (DRIVER_VERIFIER_DMA_VIOLATION).
That commit avoids triggering any DMA attempts during the shutdown of
the xHCI controller itself. However, this is not a complete solution
since any attached and opened USB device (e.g. a USB NIC) may
asynchronously trigger DMA attempts that happen to occur after bus
mastering has been disabled but before we reset the xHCI controller.
Avoid this problem by installing our own ExitBootServices notification
event at TPL_NOTIFY, thereby causing it to be invoked before the
firmware's own ExitBootServices notification event that disables bus
mastering.
This unsurprisingly causes the shutdown hook itself to be invoked at
TPL_NOTIFY, which causes a fatal error when later code attempts to
raise the TPL to TPL_CALLBACK (which is a lower TPL). Work around
this problem by redefining the "internal" iPXE TPL to be variable, and
set this internal TPL to TPL_NOTIFY when the shutdown hook is invoked.
Avoid calling into an underlying SNP protocol instance from within our
shutdown hook at TPL_NOTIFY, since the underlying SNP driver may
attempt to raise the TPL to TPL_CALLBACK (which would cause a fatal
error). Failing to shut down the underlying SNP device is safe to do
since the underlying device must, in any case, have installed its own
ExitBootServices hook if any shutdown actions are required.
Reported-by: Andreas Hammarskjöld <junior@2PintSoftware.com>
Tested-by: Andreas Hammarskjöld <junior@2PintSoftware.com>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
On some systems (observed with the Thunderbolt ports on a ThinkPad X1
Extreme Gen3 and a ThinkPad P53), the system firmware will disable bus
mastering on the xHCI controller and all PCI bridges at the point that
ExitBootServices() is called if the IOMMU is enabled. This leaves the
xHCI controller unable to shut down cleanly since all commands will
fail with a timeout.
Commit 85eb961 ("[xhci] Allow for permanent failure of the command
mechanism") allows us to detect that this has happened and respond
cleanly. However, some unidentified hardware component (either the
xHCI controller or one of the PCI bridges) seems to manage to enqueue
the attempted DMA operation and eventually complete it after the
operating system kernel has reenabled bus mastering. This results in
a DMA operation to an area of memory that the hardware is no longer
permitted to access. On Windows with the Driver Verifier enabled,
this will result in a STOP 0xE6 (DRIVER_VERIFIER_DMA_VIOLATION).
Work around this problem by detecting when bus mastering has been
disabled, and immediately failing the device to avoid initiating any
further DMA attempts.
Reported-by: Andreas Hammarskjöld <junior@2PintSoftware.com>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Some xHCI controllers (observed with the Thunderbolt ports on a
ThinkPad X1 Extreme Gen3 and a ThinkPad P53) seem to suffer a
catastrophic failure at the point that ExitBootServices() is called if
the IOMMU is enabled. The symptoms appear to be consistent with
another UEFI driver (e.g. the IOMMU driver, or the Thunderbolt driver)
having torn down the DMA mappings, leaving the xHCI controller unable
to write to host memory. The observable effect is that all commands
fail with a timeout, and attempts to abort command execution similarly
fail since the xHCI controller is unable to report the abort
completion.
Check for failure to abort a command, and respond by performing a full
device reset (as recommended by the xHCI specification) and by marking
the device as permanently failed.
Reported-by: Andreas Hammarskjöld <junior@2PintSoftware.com>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Use the "system MAC address" provided within the DSDT/SSDT if such an
address is available and has not already been assigned to a network
device.
Tested-by: Andreas Hammarskjöld <junior@2PintSoftware.com>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The RTL8211B seems to have a bug that prevents the link from coming up
unless the MII_MMD_DATA register is cleared.
The Linux kernel driver applies this workaround (in rtl8211b_resume())
only to the specific RTL8211B PHY model, along with a matching
workaround to set bit 9 of MII_MMD_DATA when suspending the PHY.
Since we have no need to ever suspend the PHY, and since writing a
zero ought to be harmless, we just clear the register unconditionally.
Debugged-by: Nikolay Pertsev <nikolay.p@cos.flag.org>
Tested-by: Nikolay Pertsev <nikolay.p@cos.flag.org>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Use hexadecimal values instead of macros in PCI_ROM entries so Perl
script can parse them correctly. Move PCI_ROM entries from header
file to C file. Integrate bnxt_vf_nics array into PCI_ROM entries by
introducing BNXT_FLAG_PCI_VF flag into driver_data field. Add
whitespaces in PCI_ROM entries for style consistency.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The use of jumbo frames for the Xen netfront virtual NIC requires the
use of scatter-gather ("feature-sg"), with the receive descriptor ring
becoming a list of page-sized buffers and the backend using as many
page buffers as required for each packet.
Since iPXE's abstraction of an I/O buffer does not include any sort of
scatter-gather list, this requires an extra allocation and copy on the
receive datapath for any packet that spans more than a single page.
This support is required in order to successfully boot an AWS EC2
virtual machine (with non-enhanced networking) via iSCSI if jumbo
frames are enabled, since the netback driver used in EC2 seems not to
allow "feature-sg" to be renegotiated once the Linux kernel driver
takes over.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
There is no method for obtaining the number of PCI buses when using
PCIAPI_DIRECT, and we therefore currently scan all possible bus
numbers. This can cause a several-second startup delay in some
virtualised environments, since PCI configuration space access will
necessarily require the involvement of the hypervisor.
Ameliorate this situation by defaulting to scanning only a single bus,
and expanding the number of PCI buses to accommodate any subordinate
buses that are detected during enumeration.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Adding this missing identifier allows the X557-AT2 chipset seen on (at
least) Super Micro A2SDI-H-TF motherboards to function with iPXE.
Signed-off-by: Tyler J. Stachecki <stachecki.tyler@gmail.com>
Allow arbitrary settings to be specified on the Linux command line.
For example:
./bin-x86_64-linux/slirp.linux \
--net slirp,testserver=qa-test.ipxe.org
This can be useful when using the Linux userspace build to test
embedded scripts, since it allows arbitrary parameters to be passed
directly on the command line.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Add a driver using libslirp to provide a virtual network interface
without requiring root permissions on the host. This simplifies the
process of running iPXE as a Linux userspace application with network
access. For example:
make bin-x86_64-linux/slirp.linux
./bin-x86_64-linux/slirp.linux --net slirp
libslirp will provide a built-in emulated DHCP server and NAT router.
Settings such as the boot filename may be controlled via command-line
options. For example:
./bin-x86_64-linux/slirp.linux \
--net slirp,filename=http://192.168.0.1/boot.ipxe
Signed-off-by: Michael Brown <mcb30@ipxe.org>
When building as a Linux userspace application, iPXE currently
implements its own system calls to the host kernel rather than relying
on the host's C library. The output binary is statically linked and
has no external dependencies.
This matches the general philosophy of other platforms on which iPXE
runs, since there are no external libraries available on either BIOS
or UEFI bare metal. However, it would be useful for the Linux
userspace application to be able to link against host libraries such
as libslirp.
Modify the build process to perform a two-stage link: first picking
out the requested objects in the usual way from blib.a but with
relocations left present, then linking again with a helper object to
create a standard hosted application. The helper object provides the
standard main() entry point and wrappers for the Linux system calls
required by the iPXE Linux drivers and interface code.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The GCC11 compiler pointed out something that apparently no previous
compiler noticed: in ath5k_eeprom_pread_turbo_modes, local variable
val is used uninitialized. From what I can see, the code is just
missing an initial AR5K_EEPROM_READ. Add it right before the switch
statement.
Signed-off-by: Bruce Rogers <brogers@suse.com>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The Linux and FreeBSD drivers for the (totally undocumented) ENA
adapters use a two-phase reset mechanism: first set ENA_CTRL.RESET and
wait for this to be reflected in ENA_STAT.RESET, then clear
ENA_CTRL.RESET and again wait for it to be reflected in
ENA_STAT.RESET.
The iPXE driver currently assumes a self-clearing reset mechanism,
which appeared to work at the time that the driver was created but
seems no longer to function, at least on the t3.nano and t3a.nano
instance types found in eu-west-1.
Switch to a simplified version of the two-phase reset mechanism as
used by Linux and FreeBSD.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
After a ton of tedious work, I am pleased to finally introduce full
support for ConnectX-3 cards in iPXE!
The work has been done by finding all publicly available versions of
the Mellanox Flexboot sources, cleaning them up, synthesizing a git
history from them, cleaning out non-significant changes, and
correlating with the iPXE upstream git history.
After this, a proof-of-concept diff was produced, that allowed iPXE to
be compiled with rudimentary ConnectX-3 support. This diff was over
10k lines, and contained many changes that were not part of the core
driver.
Special thanks to Michael Brown <mcb30@ipxe.org> for answering my
barrage of questions, and helping brainstorm the development along the
way.
Signed-off-by: Christian Iversen <ci@iversenit.dk>
The CQE length field will not be valid for a completion in error.
Avoid parsing the length field and just call the completion handler
directly.
In debug builds, also dump the queue pair context to allow for
inspection of the error.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Check for reset completion by waiting for the device to respond to PCI
configuration cycles, as documented in the Programmer's Reference
Manual. On the original ConnectX HCA, this reduces the time spent on
reset from 1000ms down to 1ms.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
When auto-detecting the initial port type, the Hermon driver will spam
the debug output without hesitation. Add a short delay in each
iteration to fix this.
Signed-off-by: Christian Iversen <ci@iversenit.dk>
Inspired by Flexboot, the function hermon_event_port_mgmnt_change() is
added to handle the HERMON_EV_PORT_MGMNT_CHANGE event type, which
updates the Infiniband subsystem.
Signed-off-by: Christian Iversen <ci@iversenit.dk>
Modified-by: Michael Brown <mcb30@ipxe.org>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Hermon Ethernet work queues have more RX than TX entries, unlike most
other drivers. This is possibly the source of some stochastic
deadlocks previously experienced with this driver.
Update the sizes to be in line with other drivers, and make them
slightly larger for better performance. These new queue sizes have
been found to work well with ConnectX-3 hardware.
Signed-off-by: Christian Iversen <ci@iversenit.dk>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The programming documentation states that the reset magic value is
"0x00000001 (Big Endian)", and the current code matches this by using
the value 0x01000000 for the implicitly little-endian writel().
Inspection of the FlexBoot source code reveals an exciting variety of
reset values, some suggestive of confusion around endianness.
Experimentation suggests that the value 0x01000001 works reliably
across a wide range of hardware.
Debugged-by: Christian Iversen <ci@iversenit.dk>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Some older versions of the hardware (and/or firmware) do not report an
event when an Infiniband link reaches the INIT state. The driver
works around this missing event by calling ib_smc_update() on each
event queue poll while the link is in the DOWN state.
Commit 6cb12ee ("[hermon] Increase polling rate for command
completions") addressed this by speeding up the time taken to issue
each command invoked by ib_smc_update(). Experimentation shows that
the impact is still significant: for example, in a situation where an
unplugged port is opened, the throughput on the other port can be
reduced by over 99%.
Fix by throttling the rate at which link polling is attempted.
Debugged-by: Christian Iversen <ci@iversenit.dk>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
This change is ported from Flexboot sources. When stopping a Hermon
device, perform hermon_unmap_mpt() which runs HERMON_HCR_HW2SW_MPT to
bring the Memory Protection Table (MPT) back to software control.
Signed-off-by: Christian Iversen <ci@iversenit.dk>
Modified-by: Michael Brown <mcb30@ipxe.org>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The eIPoIB local Ethernet MAC is currently constructed from the port
GUID. Given a base GUID/MAC value of N, Mellanox seems to populate:
Node GUID: N + 0
Port 1 GUID: N + 1
Port 2 GUID: N + 2
and
Port 1 MAC: N + 0
Port 2 MAC: N + 1
This causes a duplicate local MAC address when port 1 is configured as
Infiniband and port 2 as Ethernet, since both will derive their MAC
address as (N + 1).
Fix by using the port's Ethernet MAC as the eIPoIB local EMAC. This
is a behavioural change that could potentially break configurations
that rely on the local EMAC value, such as a DHCP server relying on
the chaddr field for DHCP reservations.
Signed-off-by: Christian Iversen <ci@iversenit.dk>
Modified-by: Michael Brown <mcb30@ipxe.org>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Some older versions of the hardware (and/or firmware) do not report an
event when an Infiniband link reaches the INIT state. The driver
works around this missing event by calling ib_smc_update() on each
event queue poll while the link is in the DOWN state. This results in
a very large number of commands being issued while any open Infiniband
link is in the DOWN state (e.g. unplugged), to the point that the 1ms
delay from waiting for each command to complete will noticeably affect
responsiveness.
Fix by decreasing the command completion polling delay from 1ms to
10us.
Signed-off-by: Christian Iversen <ci@iversenit.dk>
Modified-by: Michael Brown <mcb30@ipxe.org>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Add hermon_dump_eqctx() for dumping the event queue context and
hermon_dump_eqes() for dumping any unconsumed event queue entries.
Originally-implemented-by: Christian Iversen <ci@iversenit.dk>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Some commands (particularly in relation to device initialization) can
occasionally take longer than 2 seconds, and the Mellanox documentation
recommends a 10 second timeout.
Signed-off-by: Christian Iversen <ci@iversenit.dk>
Require drivers to report the total number of Infiniband ports. This
is necessary to report the correct number of ports on devices with
dynamic port types.
For example, dual-port Mellanox cards configured for (eth, ib) would
be rejected by the subnet manager, because they report using "port 2,
out of 1".
Signed-off-by: Christian Iversen <ci@iversenit.dk>
Some versions of gcc (observed with gcc 9.3.0 on NixOS Linux) produce
a spurious warning about an out-of-bounds array access for the
isa_extra_probe_addrs[] array.
Work around this compiler bug by redefining the array index as a
signed long, which seems to somehow avoid this spurious warning.
Debugged-by: Manuel Mendez <mmendez534@gmail.com>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Ensure that any command failure messages are followed up with an error
message indicating what the failed command was attempting to perform.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The xHCI driver can handle only a single command TRB in progress at
any one time. Immediately fail any attempts to issue concurrent
commands (which should not occur in normal operation).
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Email from solarflare.com will stop working, so update those. Remove
email for Shradha Shah, as she is not involved with this any more.
Update copyright notices for files touched.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
USB tethering via an iPhone is unreasonably complicated due to the
requirement to perform a pairing operation that involves establishing
a TLS session over a completely unrelated USB function that speaks a
protocol that is almost, but not quite, entirely unlike TCP.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Include a potential DMA mapping within the definition of an I/O
buffer, and move all I/O buffer DMA mapping functions from dma.h to
iobuf.h. This avoids the need for drivers to maintain a separate list
of DMA mappings for each I/O buffer that they may handle.
Network device drivers typically do not keep track of transmit I/O
buffers, since the network device core already maintains a transmit
queue. Drivers will typically call netdev_tx_complete_next() to
complete a transmission without first obtaining the relevant I/O
buffer pointer (and will rely on the network device core automatically
cancelling any pending transmissions when the device is closed).
To allow this driver design approach to be retained, update the
netdev_tx_complete() family of functions to automatically perform the
DMA unmapping operation if required. For symmetry, also update the
netdev_rx() family of functions to behave the same way.
As a further convenience for drivers, allow the network device core to
automatically perform DMA mapping on the transmit datapath before
calling the driver's transmit() method. This avoids the need to
introduce a mapping error handling code path into the typically
error-free transmit methods.
With these changes, the modifications required to update a typical
network device driver to use the new DMA API are fairly minimal:
- Allocate and free descriptor rings and similar coherent structures
using dma_alloc()/dma_free() rather than malloc_phys()/free_phys()
- Allocate and free receive buffers using alloc_rx_iob()/free_rx_iob()
rather than alloc_iob()/free_iob()
- Calculate DMA addresses using dma() or iob_dma() rather than
virt_to_bus()
- Set a 64-bit DMA mask if needed using dma_set_mask_64bit() and
thereafter eliminate checks on DMA address ranges
- Either record the DMA device in netdev->dma, or call iob_map_tx() as
part of the transmit() method
- Ensure that debug messages use virt_to_phys() when displaying
"hardware" addresses
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Redefine the value stored within a DMA mapping to be the offset
between physical addresses and DMA addresses within the mapped region.
Provide a dma() wrapper function to calculate the DMA address for any
pointer within a mapped region, thereby simplifying the use cases when
a device needs to be given addresses other than the region start
address.
On a platform using the "flat" DMA implementation the DMA offset for
any mapped region is always zero, with the result that dma_map() can
be optimised away completely and dma() reduces to a straightforward
call to virt_to_phys().
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>
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>
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>
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>
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>
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>
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>
A zero divisor will currently lead to a 16-bit integer overflow when
calculating the transmit padding, and a potential division by zero if
assertions are enabled.
Avoid these problems by treating a divisor value of zero as equivalent
to a divisor value of one (i.e. no alignment requirements).
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The current error handling mechanism defers the endpoint reset until
the next use of the endpoint, on the basis that errors are detected
during completions and completion handling should not recursively call
usb_poll().
In the case of usb_control(), we are already at the level that calls
usb_poll() and can therefore safely perform the endpoint reset
immediately. This has no impact on functionality, but does make
debugging traces easier to read since the reset will appear
immediately after the causative error.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
We currently use a heuristic to determine whether or not to request
cable detection in PXE_OPCODE_INITIALIZE, based on the need to work
around a known Emulex driver bug (see commit c0b61ba "[efi] Work
around bugs in Emulex NII driver") and the need to accommodate links
that are legitimately slow to come up (see commit 6324227 "[efi] Skip
cable detection at initialisation where possible").
This heuristic appears to fail with newer Emulex drivers. Attempt to
support all known drivers (past and present) by first attempting
initialisation with cable detection, then falling back to attempting
initialisation without cable detection.
Reported-by: Kwang Woo Lee <kwleeyh@gmail.com>
Tested-by: Kwang Woo Lee <kwleeyh@gmail.com>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
A failure in device registration (e.g. due to a device with malformed
descriptors) will currently result in the port being disabled as part
of the error path. This in turn causes the hardware to detect the
device as newly connected, leading to an endless loop of failed device
registrations.
Fix by leaving the port enabled in the case of a registration failure.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
When connected to a USB3 port, the AX88179 seems to have an
approximately 50% chance of producing a USB transaction error on each
of its three endpoints after being closed and reopened. The root
cause is unclear, but rewriting the USB device configuration value
seems to clear whatever internal error state has accumulated.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Experimentation shows that the existing 20ms delay is insufficient,
and often results in device detection being deferred until after iPXE
has completed startup.
Fix by increasing the delay to 100ms.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The driver-private data for root hubs is already set immediately after
allocating the USB bus. There seems to be no reason to set it again
when opening the root hub.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The "disabled" port states for USB2 and USB3 are not directly
equivalent. In particular, a disabled USB3 port will not detect new
device connections. The result is that a USB3 device disconnected
from and reconnected to an xHCI root hub port will end up reconnecting
as a USB2 device.
Fix by setting the link state to RxDetect after disabling the port, as
is already done during initialisation.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The USB3 specification removes PORT_ENABLE from the list of features
that may be cleared via a CLEAR_FEATURE request. Experimentation
shows that omitting the attempt to clear PORT_ENABLE seems to result
in the correct hotplug behaviour.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Resetting the host endpoint may immediately restart any pending
transfers for that endpoint. If the device endpoint halt has not yet
been cleared, then this will probably result in a second failed
transfer.
This second failure may be detected within usb_endpoint_reset() while
waiting for usb_clear_feature() to complete. The endpoint will
subsequently be removed from the list of halted endpoints, causing the
second failure to be effectively ignored and leaving the host endpoint
in a permanently halted state.
Fix by deferring the host endpoint reset until after the device
endpoint is ready to accept new transfers.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The ASIX USB NICs are capable of autodetecting the Ethernet link speed
and reporting it via PLSR but will not automatically update the
relevant GM and PS bits in MSR. The result is that a non-gigabit link
will fail to send or receive any packets.
The interrupt endpoint used to report link state includes the values
of the PHY BMSR and LPA registers. These are not sufficient to
differentiate between 100Mbps and 1000Mbps, since the LPA_NPAGE bit
does not necessarily indicate that the link partner is advertising
1000Mbps.
Extend axge_check_link() to write the MSR value based on the link
speed read from PLSR, and simplify the interrupt endpoint handler to
merely trigger a call to axge_check_link().
Signed-off-by: Michael Brown <mcb30@ipxe.org>
GCC 10 emits warnings for implicit conversions of enumerated types.
The flexboot_nodnic code defines nodnic_queue_pair_type with values
identical to those of ib_queue_pair_type, and implicitly casts between
them. Add an explicit cast to fix the warning.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
GCC 10 produces a spurious warning about an out-of-bounds array access
for the unsized raw dword array in union intelvf_msg.
Avoid the warning by embedding the zero-length array within a struct.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Some devices (observed with a Getac RX10 tablet and docking station
containing an embedded AX88179 USB NIC) seem to be capable of
detecting link state only during the call to Initialize(), and will
occasionally erroneously report that the link is down for the first
few such calls.
Work around these devices by retrying the Initialize() call multiple
times, terminating early if a link is detected. The eventual absence
of a link is treated as a non-fatal error, since it is entirely
possible that the link really is down.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
According to UEFI specification 2.8 p 24.1 we must set the
EFI_SIMPLE_NETWORK_RECEIVE_MULTICAST bit in the "Disable" mask, when
"ResetMCastFilter" is TRUE.
Signed-off-by: Ignat Korchagin <ignat@cloudflare.com>
Split-by: Michael Brown <mcb30@ipxe.org>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Currently, if the SNP driver for whatever reason fails to enable
receive filters for multicast frames, it falls back to enabling just
unicast and broadcast filters. This breaks some IPv6 functionality as
the network card does not respond to neighbour solicitation requests.
Some cards refuse to enable EFI_SIMPLE_NETWORK_RECEIVE_MULTICAST, but
do support enabling EFI_SIMPLE_NETWORK_RECEIVE_PROMISCUOUS_MULTICAST,
so try it before falling back to just unicast+broadcast.
Signed-off-by: Ignat Korchagin <ignat@cloudflare.com>
Split-by: Michael Brown <mcb30@ipxe.org>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
On devices with no EEPROM or OTP, the MAC_CR register defaults to not
using automatic link speed detection, with the result that no packets
are successfully sent or received.
Fix by always enabling automatic speed and duplex detection, since
iPXE provides no mechanism for manual configuration of either link
speed or duplex.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
On at least some platforms (observed with a Raspberry Pi), any attempt
to perform USB transfers via EFI_USB_IO_PROTOCOL during EFI shutdown
will lock up the system. This is quite probably due to the already
documented failure of all EFI timers when ExitBootServices() is
called: see e.g. commit 5cf5ffea2 "[efi] Work around temporal anomaly
encountered during ExitBootServices()".
Work around this problem by refusing to poll endpoints if shutdown is
in progress, and by immediately failing any attempts to enqueue new
transfers.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The USB core reuses the I/O buffer space occupied by the USB setup
packet to hold the completion status for message transfers, assuming
that the message() method will always strip the setup packet before
returning. This assumption is correct for all of the hardware
controller drivers (XHCI, EHCI, and UHCI), since these drivers are
able to enqueue the transfer as a separate action from waiting for the
transfer to complete.
The EFI_USB_IO_PROTOCOL does not allow us to separate actions in this
way: there is only a single blocking method that both enqueues and
waits for completion. Our usbio driver therefore currently defers
stripping the setup packet until the control endpoint is polled.
This causes a bug if a message transfer is enqueued but never polled
and is subsequently cancelled, since the cancellation will be reported
with the I/O buffer still containing the setup packet. This breaks
the assumption that the setup packet has been stripped, and triggers
an assertion failure in usb_control_complete().
Fix by always stripping the setup packet in usbio_endpoint_message(),
and adjusting usbio_control_poll() to match.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The physical function defaults to operating in "PXE mode" after a
power-on reset. In this mode, receive descriptors are fetched and
written back as single descriptors. In normal (non-PXE mode)
operation, receive descriptors are fetched and written back only as
complete cachelines unless an interrupt is raised.
There is no way to return to PXE mode from non-PXE mode, and there is
no way for the virtual function driver to operate in PXE mode.
Choose to operate in non-PXE mode. This requires us to trick the
hardware into believing that it is raising an interrupt, so that it
will not defer writing back receive descriptors until a complete
cacheline (i.e. four packets) have been consumed. We do so by
configuring the hardware to use MSI-X with a dummy target location in
place of the usual APIC register.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The virtual function driver will use the same transmit and receive
descriptor ring structures, but will not itself construct and program
the ring context. Split out ring creation and destruction from the
programming of the ring context, to allow code to be shared between
physical and virtual function drivers.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The virtual function transmit and receive ring tail register offsets
do not match those of the physical function. Allow the tail register
offsets to be specified separately.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The physical function driver does not allow the virtual function to
request the use of 16-byte receive descriptors. Switch to using
32-byte receive descriptors.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Provide a weak stub function for handling the "send to VF" event used
for communications between the physical and virtual function drivers.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The "send to PF" and "send to VF" admin queue descriptors (ab)use the
cookie field to hold the extended opcode and return code values.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
A virtual function reset is triggered via an admin queue command and
will reset the admin queue configuration registers. Allow the admin
queues to be reinitialised after such a reset, without requiring the
overhead (and potential failure paths) of freeing and reallocating the
queues.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
We currently use a single data buffer shared between all admin queue
descriptors. This works for the physical function driver since we
have at most one command in progress and only a single event (which
does not use a data buffer).
The communication path between the physical and virtual function
drivers uses the event data buffer, and there is no way to prevent a
solicited event (i.e. a response to a request) from being overwritten
by an unsolicited event (e.g. a link status change).
Provide individual data buffers for each admin event queue descriptor
(and for each admin command queue descriptor, for the sake of
consistency).
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The register map for the virtual functions appears to have been
constructed using a random number generator.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The physical function driver does not allow the virtual function to
request that VLAN tags are left unstripped. Extract and use the VLAN
tag from the receive descriptor if present.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The Hermon driver uses vlan_find() to identify the appropriate VLAN
device for packets that are received with the VLAN tag already
stripped out by the hardware. Generalise this capability and expose
it for use by other network card drivers.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The Intel 40 Gigabit Ethernet virtual functions support only MSI-X
interrupts, and will write back completed interrupt descriptors only
when the device attempts to raise an interrupt (or when a complete
cacheline of receive descriptors has been completed).
We cannot actually use MSI-X interrupts within iPXE, since we never
have ownership of the APIC. However, an MSI-X interrupt is
fundamentally just a DMA write of a single dword to an arbitrary
address. We can therefore configure the device to "raise" an
interrupt by writing a meaningless value to an otherwise unused memory
location: this is sufficient to trigger the receive descriptor
writeback logic.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
The first adapters in this family are X2522-10, X2522-25, X2541 and
X2542.
These no longer use PCI BAR 0 for I/O, but use that for memory. In
other words, BAR 2 on SFN8xxx adapters now becomes BAR 0.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Devices that support jumbo frames will currently default to the
largest possible MTU. This assumption is valid for virtual adapters
such as virtio-net, where the MTU must have been configured by a
system administrator, but is unsafe in the general case of a physical
adapter.
Default to the standard Ethernet MTU, unless explicitly overridden
either by the driver or via the ${netX/mtu} setting.
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Add the function mii_find() in order to locate the PHY address.
Signed-off-by: Sylvie Barlow <sylvie.c.barlow@gmail.com>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
We currently have no generic concept of a PHY address, since all
existing implementations simply hardcode the PHY address within the
MII access methods.
A bit-bashing MII interface will need to be provided with an explicit
PHY address in order to generate the correct waveform. Allow for this
by separating out the concept of a MII device (i.e. a specific PHY
address attached to a particular MII interface).
Signed-off-by: Michael Brown <mcb30@ipxe.org>
This is required to work around a bug in some firmware versions.
Signed-off-by: Ameer Mahagneh <ameerm@mellanox.com>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Some drivers are known to call the optional Map_Mem() callback without
first checking that the callback exists. Provide a usable basic
implementation of Map_Mem() along with the other callbacks that become
mandatory if Map_Mem() is provided.
Note that in theory the PCI I/O protocol is allowed to require
multiple calls to Map(), with each call handling only a subset of the
overall mapped range. However, the reference implementation in EDK2
assumes that a single Map() will always suffice, so we can probably
make the same simplifying assumption here.
Tested with the Intel E3522X2.EFI driver (which, incidentally, fails
to cleanly remove one of its mappings).
Originally-implemented-by: Maor Dickman <maord@mellanox.com>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Remove the global variable shomron_nodnic_supported, since it may have
different values for different PCI devices.
Originally-fixed-by: Mohammed Taha <mohammedt@mellanox.com>
Signed-off-by: Michael Brown <mcb30@ipxe.org>
Michael Brown
Benedikt Braunger
Aaron Young
Joseph
Christian Nilsson
Tyler J. Stachecki
Bruce Rogers
Christian Iversen
Manuel Mendez
Martin Habets
Mohammed Taha
Rob Taglang
Ben Bleything
Andrew Grasso
LaDerrick H
Ignat Korchagin
Mohammed
Steven Haber
Sylvie Barlow
Rob Taglang
Ameer Mahagneh