There is diagnostic value in being able to disambiguate between the
various reasons why an IB CM has rejected a connection attempt. In
particular, reason 8 "invalid service ID" can be used to identify an
incorrect SRP service_id root-path component, and reason 28 "consumer
reject" corresponds to a genuine SRP login rejection IU, which can be
passed up to the SRP layer.
For rejection reasons other than "consumer reject", we should not pass
through the private data, since it is most likely generated by the CM
without any protocol-specific knowledge.
Generate errors within individual MAD transaction consumers such as
ib_pathrec.c and ib_mcast.c, rather than within ib_mi.c. This allows
for more meaningful error messages to eventually be displayed to the
user.
SRP is the SCSI RDMA Protocol. It allows for a method of SAN booting
whereby the target is responsible for reading and writing data using
Remote DMA directly to the initiator's memory. The software initiator
merely sends and receives SCSI commands; it never has to touch the
actual data.
The minimal-surprise behaviour, when no explicit SRP initiator device
is specified, will probably be to use the most recently opened
Infiniband device. This matches our behaviour with using the most
recently opened net device for PXE, iSCSI, AoE, NBI, etc.
SRP over Infiniband uses a protocol whereby data is sent via a
combination of the CM private data fields and the RC queue pair
itself. This seems sufficiently generic that it's worth having
available as a separate protocol.
We will terminate our transaction as soon as we receive the first CM
REP, since that provides all the state that we need. However, the
peer may resend the REP if it didn't see our RTU, and if we don't
respond with another RTU we risk being disconnected. (This protocol
appears not to handle retries gracefully.)
Fix by adding a management agent that will listen for these duplicate
REPs and send back an RTU.
When a probe found no results, the list head of beacons would not be
freed, leaking 16 bytes of memory per probe.
Signed-off-by: Michael Brown <mcb30@etherboot.org>
Some cards (such as ath5k) always need to tune to a particular channel
when they are reset; the reset may happen upon open(), which is before
the channels array would be set up (in prepare_probe()). Avoid tuning
the card to an inconsistent state by copying the hardware
supported-channels array to the 802.11 device's allowable-channels
array even before channels are "properly" set up.
Signed-off-by: Michael Brown <mcb30@etherboot.org>
The prior net80211 model of physical-layer behavior for drivers was
overly simplistic and limited the drivers that could be written. To
be more flexible, split the driver-provided list of supported rates by
band, and add a means for specifying a list of supported channels.
Allow drivers to specify a hardware channel value that will be tied to
uses of the channel.
Expose net80211_duration() to drivers, and make the rate it uses in
its computations configurable, so that it can be used in calculating
durations that must be set in hardware for ACK and CTS packets. Add
net80211_cts_duration() for the common case of calculating the
duration for a CTS packet.
Signed-off-by: Michael Brown <mcb30@etherboot.org>
A management interface is the component through which both local and
remote management agents are accessed.
This new implementation of a management interface allows for the user
to react to timed-out transactions, and also allows for cancellation
of in-progress transactions.
The IBA specification refers to management "interfaces" and "agents".
The interface is the component that connects to the queue pair and
sends and receives MADs; the agent is the component that constructs
the reply to the MAD.
Rename the IB_{QPN,QKEY,QPT} constants as a first step towards making
this separation in gPXE.
In several places, we currently use size_t to represent a difference
between TCP sequence numbers. This can cause compiler warnings
relating to printf format specifiers, since the result of
(uint32_t+size_t) may be an unsigned long on some compilers.
Fix by using uint32_t for all variables that represent a difference
between TCP sequence numbers.
Tested-by: Joshua Oreman <oremanj@xenon.get-linux.org>
This is required for all modern 802.11 devices, and allows drivers
to be written for them with minimally more effort than is required
for a wired NIC.
Signed-off-by: Michael Brown <mcb30@etherboot.org>
Modified-by: Michael Brown <mcb30@etherboot.org>
Queue pairs are now assumed to be created in the INIT state, with a
call to ib_modify_qp() required to bring the queue pair to the RTS
state.
ib_modify_qp() no longer takes a modification list; callers should
modify the relevant queue pair parameters (e.g. qkey) directly and
then call ib_modify_qp() to synchronise the changes to the hardware.
The packet sequence number is now a property of the queue pair, rather
than of the device.
Each queue pair may have an associated address vector. For RC queue
pairs, this is the address vector that will be programmed in to the
hardware as the remote address. For UD queue pairs, it will be used
as the default address vector if none is supplied to ib_post_send().
Now that MAD handlers no longer return a status code, we can allow
them to return a pointer to a MAD structure if and only if they want
to send a response. This provides a more natural and flexible
approach than using a "response method" field within the handler's
descriptor.
MAD handlers have to set the status fields within the MAD itself
anyway, in order to provide a meaningful response MAD; the additional
gPXE return status code is just noise.
Note that we probably don't need to ever explicitly set the status to
IB_MGMT_STATUS_OK, since it should already have this value from the
request. (By not explicitly setting the status in this way, we can
safely have ib_sma_set_xxx() call ib_sma_get_xxx() in order to
generate the GetResponse MAD without worrying that ib_sma_get_xxx()
will clear any error status set by ib_sma_set_xxx().)
Most IB hardware seems not to allow allocation of the genuine QPNs 0
and 1, so allow for the externally-visible QPN (as constructed and
parsed by ib_packet, where used) to differ from the real
hardware-allocated QPN.
The queue key is stored as a property of the queue pair, and so can
optionally be added by the Infiniband core at the time of calling
ib_post_send(), rather than always having to be specified by the
caller.
This allows IPoIB to avoid explicitly keeping track of the data queue
key.
Generalise the subnet management agent into a general management agent
capable of sending and responding to MADs, including support for
retransmissions as necessary.
Currently, all Infiniband users must create a process for polling
their completion queues (or rely on a regular hook such as
netdev_poll() in ipoib.c).
Move instead to a model whereby the Infiniband core maintains a single
process calling ib_poll_eq(), and polling the event queue triggers
polls of the applicable completion queues. (At present, the
Infiniband core simply polls all of the device's completion queues.)
Polling a completion queue will now implicitly refill all attached
receive work queues; this is analogous to the way that netdev_poll()
implicitly refills the RX ring.
Infiniband users no longer need to create a process just to poll their
completion queues and refill their receive rings.
IPoIB and the SMA have separate constants for the packet size to be
used to I/O buffer allocations. Merge these into the single
IB_MAX_PAYLOAD_SIZE constant.
(Various other points in the Infiniband stack have hard-coded
assumptions of a 2048-byte payload; we don't currently support
variable MTUs.)
IPoIB has a link-layer broadcast address that varies according to the
partition key. We currently go through several contortions to pretend
that the link-layer address is a fixed constant; by making the
broadcast address a property of the network device rather than the
link-layer protocol it will be possible to simplify IPoIB's broadcast
handling.
Move the icky call to step() from aoe.c to ata.c; this takes it at
least one step further away from where it really doesn't belong.
Unfortunately, AoE has the ugly aoe_discover() mechanism which means
that we still have a step() loop in aoe.c for now; this needs to be
replaced at some future point.
Expand the NETDEV_LINK_UP bit into a link_rc status code field,
allowing specific reasons for link failure to be reported via
"ifstat".
Originally-authored-by: Joshua Oreman <oremanj@rwcr.net>
Commit 558c1a4 ("[tcp] Improve robustness in the presence of duplicated
received packets") introduced a regression in that an old duplicate
ACK received while in the ESTABLISHED state would pass through normal
ACK processing, including updating tcp->snd_seq.
Fix by ensuring that ACK processing ignores all duplicate ACKs.
All TCP errors or unusual events should now generate a debugging
message at DBGLVL_LOG, with enough information (SEQ and ACK numbers)
to be able to identify the corresponding packet (or missing packet) in
a network trace from the remote end.
This makes it possible to leave TCP debugging enabled in order to see
interesting TCP events, without flooding the console with at least one
message per packet.
In order to construct outgoing link-layer frames or parse incoming
ones properly, some protocols (such as 802.11) need more state than is
available in the existing variables passed to the link-layer protocol
handlers. To remedy this, add struct net_device *netdev as the first
argument to each of these functions, so that more information can be
fetched from the link layer-private part of the network device.
Updated all three call sites (netdevice.c, efi_snp.c, pxe_undi.c) and
both implementations (ethernet.c, ipoib.c) of ll_protocol to use the
new argument.
Signed-off-by: Michael Brown <mcb30@etherboot.org>
gPXE responds to duplicated ACKs with an immediate retransmission,
which can lead to a sorceror's apprentice syndrome. It also responds
to out-of-range (or old duplicate) ACKs with a RST, which can cause
valid connections to be dropped.
Fix the sorceror's apprentice syndrome by leaving the retransmission
timer running (and so inhibiting the immediate retransmission) when we
receive a potential duplicate ACK. This seems to match the behaviour
of Linux observed via wireshark traces.
Fix the RST issue by sending RST only on out-of-range ACKs that occur
before the connection is fully established, as per RFC 793.
These problems were exposed during development of the 802.11 wireless
link layer; the 802.11 protocol has a failure mode that can easily
cause duplicated packets. The fixes were tested in a controlled way
by faking large numbers of duplicated packets in the rtl8139 driver.
Originally-fixed-by: Joshua Oreman <oremanj@rwcr.net>
If the ProxyDHCPOFFER already includes PXE options (i.e. option 60 is
set to "PXEClient" and option 43 is present) then assume that the
ProxyDHCPREQUEST can be sent to port 67, rather than port 4011. This
is a reasonable assumption, since in that case the ProxyDHCP server
has already demonstrated by responding to the DHCPDISCOVER that it is
listening on port 67. (If the ProxyDHCP server were not listening on
port 67, then the standard DHCP server would have been configured to
respond with option 60 set to "PXEClient" but no option 43 present.)
The PXE specification is ambiguous on this point; the specified
behaviour covers only the cases in which option 43 is *not* present in
the ProxyDHCPOFFER. In these cases, we will continue to send the
ProxyDHCPREQUEST to port 4011.
This change is required in order to allow us to interoperate with
dnsmasq, which listens only on port 67. (dnsmasq relies on
unspecified behaviour of the Intel PXE stack, which it seems will
retain the ProxyDHCPOFFER as an options source and never issue a
ProxyDHCPREQUEST, thereby enabling dnsmasq to omit listening on port
4011.)
IBM Tivoli PXE Server 5.1.0.3 is reported to send trailing garbage
bytes at the end of the OACK packet, which causes gPXE to reject the
packet and abort the TFTP transfer.
Work around the problem by processing as much as possible of the OACK,
and treating name/value parsing errors as non-fatal.
Reported-by: Shao Miller <Shao.Miller@yrdsb.edu.on.ca>
We currently send all boot server discovery requests to port 4011.
Section 2.2.1 of the PXE spec states that boot server discovery
packets should be "sent broadcast (port 67), multicast (port 4011), or
unicast (port 4011)". Adjust our behaviour so that any boot server
discovery packets that are sent to the broadcast address are directed
to port 67 rather than port 4011.
This is required for operation with dnsmasq as a PXE server, since
dnsmasq listens only on port 67, and relies upon this (specified)
behaviour.
This change may break some setups using the (itself very broken) Linux
PXE server from kano.org.uk. This server will, in its default
configuration, listen only on port 4011. It never constructs a boot
server list (PXE_BOOT_SERVERS, option 43.8), and uses the wrong
definitions for the discovery control bits (PXE_DISCOVERY_CONTROL,
option 43.6). The upshot is that it will always instruct the client
to perform multicast and broadcast discovery only. In setups lacking
a valid multicast route on the server side, this used to work because
gPXE would eventually give up on the (non-responsive) multicast
address and send a broadcast request to port 4011, which the Linux PXE
server would respond to. Now that gPXE correctly sends this broadcast
request to port 67 instead, it is never seen by the Linux PXE server,
and the boot fails. The fix is to either (a) set up a multicast route
correctly on the server side before starting the PXE server, or (b)
edit /etc/pxe.conf to contain the server's unicast address in the
"multicast_address" field (a hack that happens to work).
Suggested-by: Simon Kelley <simon@thekelleys.org.uk>
This prevents gPXE from wasting time attempting to contact a ProxyDHCP
server on port 4011 if the DHCP response already contains the relevant
PXE options. This behaviour is hinted at (though not explicitly
specified) in the PXE spec, and seems to match what the Intel client
does.
Suggested-by: Simon Kelley <simon@thekelleys.org.uk>
Eliminate the potential for mismatches between table names and the
table entry data type by incorporating the data type into the
definition of the table, rather than specifying it explicitly in each
table accessor method.
Intel's C compiler (icc) chokes on the zero-length arrays that we
currently use as part of the mechanism for accessing linker table
entries. Abstract away the zero-length arrays, to make a port to icc
easier.
Introduce macros such as for_each_table_entry() to simplify the common
case of iterating over all entries in a linker table.
Represent table names as #defined string constants rather than
unquoted literals; this avoids visual confusion between table names
and C variable or type names, and also allows us to force a
compilation error in the event of incorrect table names.
Some firewall devices seem to regard SYN,PSH as an invalid flag
combination and reject the packet. Fix by setting PSH only if SYN is
not set.
Reported-by: DSE Incorporated <dseinc@gmail.com>
Avoid passing credentials in the iBFT that were available but not
required for login. This works around a problem in the Microsoft
iSCSI initiator, which will refuse to initiate sessions if the CHAP
password is fewer than 12 characters, even if the target ends up not
asking for CHAP authentication.
It is a programming error, not a runtime error, if we attempt to use
block ciphers with an incorrect blocksize, so use an assert() rather
than an error status return.
The various types of cryptographic algorithm are fundamentally
different, and it was probably a mistake to try to handle them via a
single common type.
pubkey_algorithm is a placeholder type for now.
The documentation in xfer.h and xfer.c does not say that the metadata
parameter is optional in calls such as xfer_deliver_iob_meta() and the
deliver_iob() method. However, some code in net/ is prepared to
accept a NULL pointer, and xfer_deliver_as_iob() passes a NULL pointer
directly to the deliver_iob() method.
Fix this mess of conflicting assumptions by making everything assume
that the metadata parameter is mandatory, and fixing
xfer_deliver_as_iob() to pass in a dummy metadata structure (as is
already done in xfer_deliver_iob()).
Various combinations of options 43.6, 43.7 and 43.8 dictate which
servers we send Boot Server Discovery requests to, and which servers
we should accept responses from. Obey these options, and remove the
explicit specification of a single Boot Server from start_pxebs() and
dependent functions.
There are many functions that take ownership of the I/O buffer they
are passed as a parameter. The caller should not retain a pointer to
the I/O buffer. Use iob_disown() to automatically nullify the
caller's pointer, e.g.:
xfer_deliver_iob ( xfer, iob_disown ( iobuf ) );
This will ensure that iobuf is set to NULL for any code after the call
to xfer_deliver_iob().
iob_disown() is currently used only in places where it simplifies the
code, by avoiding an extra line explicitly setting the I/O buffer
pointer to NULL. It should ideally be used with each call to any
function that takes ownership of an I/O buffer. (The SSA
optimisations will ensure that use of iob_disown() gets optimised away
in cases where the caller makes no further use of the I/O buffer
pointer anyway.)
If gcc ever introduces an __attribute__((free)), indicating that use
of a function argument after a function call should generate a
warning, then we should use this to identify all applicable function
call sites, and add iob_disown() as necessary.
A TFTP DATA packet with a block number of zero (representing a
negative offset within the file) could potentially cause problems.
Fixed by explicitly rejecting such packets.
Identified by Stefan Hajnoczi <stefanha@gmail.com>.
The DHCP client code now implements only the mechanism of the DHCP and
PXE Boot Server protocols. Boot Server Discovery can be initiated
manually using the "pxebs" command. The menuing code is separated out
into a user-level function on a par with boot_root_path(), and is
entered in preference to a normal filename boot if the DHCP vendor
class is "PXEClient" and the PXE boot menu option exists.
Try to qualify relative names in the DNS resolver using the DHCP Domain
Name. For example:
DHCP Domain Name: etherboot.org
(Relative) Name: www
yields:
www.etherboot.org
Only names with no dots ('.') will be modified. A name with one or more
dots is unchanged.
pxe_tftp.c assumes that the first seek on its data-transfer interface
represents the block size. Apart from being an ugly hack, this will
also screw up file size calculation for files smaller than one block.
The proper solution would be to extend the data-transfer interface to
support the reporting of stat()-like data. This is not going to
happen until the cost of adding interface methods is reduced (a fix I
have planned since June 2008).
In the meantime, abuse the xfer_window() method to return the block
size, since it is not being used for anything else and is vaguely
justifiable.
Astonishingly, having returned the incorrect TFTP blocksize via
PXENV_TFTP_OPEN for almost a year seems not to have affected any of
the test cases run during that time; this bug was found only when
someone tried running the heavily-patched version of pxegrub found in
OpenSolaris.
PXE dictates a mechanism for boot menuing, involving prompting the
user with a variable message, waiting for a predefined keypress,
displaying a boot menu, and waiting for a selection.
This breaks the currently desirable abstraction that DHCP is a process
that can happen in the background without any user interaction.
Remove the lazy assumption that ProxyDHCP == "DHCP with option 60 set
to PXEClient", and explicitly separate the notion of ProxyDHCP from
the notion of packets containing PXE options.
It is possible to configure a DHCP server to hand out PXE options
without a ProxyDHCP server present. This requires setting option 60
to "PXEClient", which will cause gPXE to attempt ProxyDHCP.
We assume in several places that dhcp->proxydhcpack is set to the
DHCPACK packet containing option 60 set to "PXEClient". When we
transition into ProxyDHCPREQUEST, set dhcp->proxydhcpack=dhcp->dhcpack
so that this assumption holds true.
We ought to rename several references to "proxydhcp" to something more
accurate, such as "pxedhcp". Treating a single DHCP response as
potentially both DHCPOFFER and ProxyDHCPOFFER does make the code
smaller, but the variable names get confusing.
Pick out the first boot menu item from the boot menu (option 43.9) and
pass it to the boot server as the boot menu item (option 43.71).
Also improve DHCP debug messages to include more details of the
packets being transmitted.
Apparently this can cause a major speedup on some iSCSI targets, which
will otherwise wait for a timer to expire before responding. It
doesn't seem to hurt other simple TCP test cases (e.g. HTTP
downloads).
Problem and solution identified by Shiva Shankar <802.11e@gmail.com>
Some PXE configurations require us to perform a third DHCP transaction
(in addition to the real DHCP transaction and the ProxyDHCP
transaction) in order to retrieve information from a "Boot Server".
This is an experimental implementation, since the actual behaviour is
not well specified in the PXE spec.
When sending to a multicast address, it may be necessary to specify
the source address explicitly, since the multicast destination address
does not provide enough information to deduce the source address via
the miniroute table.
Allow the source address specified via the data-xfer metadata to be
passed down through the TCP/IP stack to the IPv4 layer, which can use
it as a default source address.
Move all the DHCP state transition logic into a single function
dhcp_next_state(). This will make it easier to add support for PXE
Boot Servers, since it abstracts away the difference between "mark
DHCP as complete" and "transition to boot server discovery".
The Linux PXE server (http://www.kano.org.uk/projects/pxe) does not
set the server identifier in its ProxyDHCP responses. If the server
ID is missing, do not treat this as an error.
This resolves the "vague and unsettling memory" mentioned in commit
fdb8481d ("[dhcp] Verify server identifier on ProxyDHCPACKs").
Note that we already accept ProxyDHCPOFFERs without a server
identifier; they get treated as potential BOOTP packets.
There are currently four places within the codebase that use a
heuristic to guess the "boot network device", with varying degrees of
success. Add a feature to the net device core to maintain a list of
open network devices, in order of opening, and provide a function
last_opened_netdev() to retrieve the most recently opened net device.
This should do a better job than the current assortment of
guess_boot_netdev() functions.
The AoE spec does not specify that the source MAC address of a
received packet actually matches the MAC address of the AoE target.
In principle an AoE server can respond to an AoE request on any
interface available to it, which may not be an address configured to
accept AoE requests.
This issue is resolved by implementing AoE device discovery. The
purpose of AoE discovery is to find out which addresses an AoE target
can use for requests. An AoE configuration command is sent when the
AoE attach is attempted. The AoE target must respond to that
configuration query from an interface that can accept requests.
Based on a patch from Ryan Thomas <ryan@coraid.com>
This brings us in to line with Linux definitions, and also simplifies
adding x86_64 support since both platforms have 2-byte shorts, 4-byte
ints and 8-byte long longs.
This can be used with cards that require the driver to construct and
parse packet headers manually. Headers are optionally handled
out-of-line from the packet payload, since some such cards will split
received headers into a separate ring buffer.
Not all Infiniband cards have embedded subnet management agents.
Split out the code that communicates with such an embedded SMA into a
separate ib_smc.c file, and have drivers call ib_smc_update()
explicitly when they suspect that the answers given by the embedded
SMA may have changed.
Receive completion handlers now get passed an address vector
containing the information extracted from the packet headers
(including the GRH, if present), and only the payload remains in the
I/O buffer.
This breaks the symmetry between transmit and receive completions, so
remove the ib_completer_t type and use an ib_completion_queue_operations
structure instead.
Rename the "destination QPN" and "destination LID" fields in struct
ib_address_vector to reflect its new dual usage.
Since the ib_completion structure now contains only an IB status code,
("syndrome") replace it with a generic gPXE integer status code.
Avoid leaking I/O buffers in ib_destroy_qp() by completing any
outstanding work queue entries with a generic error code. This
requires the completion handlers to be available to ib_destroy_qp(),
which is done by making them static configuration parameters of the CQ
(set by ib_create_cq()) rather than being provided on each call to
ib_poll_cq().
This mimics the functionality of netdev_{tx,rx}_flush(). The netdev
flush functions would previously have been catching any I/O buffers
leaked by the IPoIB data queue (though not by the IPoIB metadata
queue).
netdev_rx_err() and netdev_tx_complete_err() get passed the error
code, but currently use it only in debug messages.
Retain error numbers and frequencey counts for up to
NETDEV_MAX_UNIQUE_ERRORS (4) different errors for each of TX and RX.
This allows the "ifstat" command to report the reasons for TX/RX
errors in most cases, even in non-debug builds.
The retry timer needs to be running as soon as we know that we are
trying to transmit a command. If transmission fails because of a
temporary error condition, then the timer will allow us to retry the
transmission later.
Settings can be constructed using a dotted-decimal notation, to allow
for access to unnamed settings. The default interpretation is as a
DHCP option number (with encapsulated options represented as
"<encapsulating option>.<encapsulated option>".
In several contexts (e.g. SMBIOS, Phantom CLP), it is useful to
interpret the dotted-decimal notation as referring to non-DHCP
options. In this case, it becomes necessary for these contexts to
ignore standard DHCP options, otherwise we end up trying to, for
example, retrieve the boot filename from SMBIOS.
Allow settings blocks to specify a "tag magic". When dotted-decimal
notation is used to construct a setting, the tag magic value of the
originating settings block will be ORed in to the tag number.
Store/fetch methods can then check for the magic number before
interpreting arbitrarily-numbered settings.
EFI requires us to be able to specify the source address for
individual transmitted packets, and to be able to extract the
destination address on received packets.
Take advantage of this to rationalise the push() and pull() methods so
that push() takes a (dest,source,proto) tuple and pull() returns a
(dest,source,proto) tuple.
Multicast hashing is an ugly overlap between network and link layers.
EFI requires us to provide access to this functionality, so move it
out of ipv4.c and expose it as a method of the link layer.
-Wformat-nonliteral is not enabled by -Wall and needs to be explicitly
specified.
Modified the few files that use nonliteral format strings to work with
this new setting in place.
Inspired by a patch from Carl Karsten <carl@personnelware.com> and an
identical patch from Rorschach <r0rschach@lavabit.com>.
The domain etherboot.org was actually registered on 2000-01-09, not
2000-09-01. (To put it another way, it was registered on 1/9/2000 (US
date format) rather than 1/9/2000 (sensible date format); this may
illuminate the cause of the error.)
"iqn.2000-09.org.etherboot:" is still valid as per RFC3720, but may be
surprising to users, so change it to something less unexpected.
Thanks to the anonymous contributor for pointing this one out.
Determine the network-layer packet type and fill it in for UNDI
clients. This is required by some NBPs such as emBoot's winBoot/i.
This change requires refactoring the link-layer portions of the
gPXE netdevice API, so that it becomes possible to strip the
link-layer header without passing the packet up the network stack.
The ProxyDHCPREQUEST is a unicast packet, so the first request will
almost always be lost due to not having the IP address in the ARP
cache. If the minimum retry time is set to one second (as per commit
ff2b6a5), then ProxyDHCP will time out and give up before managing to
successfully transmit a request.
The DHCP timers need to be reworked anyway, so this mild hack is
acceptable for now.
New min_timeout and max_timeout fields in struct retry_timer allow
users of this timer to set their own desired minimum and maximum
timeouts, without being constrained to a single global minimum and
maximum. Users of the timer can still elect to use the default global
values by leaving the min_timeout and max_timeout fields as 0.
WinPE seems to have a bug that causes it to always use the TFTP server
IP address and filename from the ProxyDHCPACK packet, even if the
ProxyDHCPACK packet doesn't exist. This causes it to end up
attempting to fetch a file such as
tftp://0.0.0.0/bootmgr.exe
If we don't have a ProxyDHCPACK to use, we pretend that it was a copy
of the DHCPACK packet. This works around the problem, and hopefully
won't surprise any NBPs.
Altiris erroneously cares about the ordering of DHCP options, and will
get confused if we don't construct them in the order it expects.
This is observed (so far) only when attempting to deploy 64-bit Win2k3.
When an error reply (not 1xx, 2xx or 3xx) was received, ftp_reply()
invoked ftp_done() to close connections, but did not return, and the
rest of code in this function could try to send commands to the closed
control connection.
Signed-off-by: Sergey Vlasov <vsu@altlinux.ru>
Based on a patch contributed by Sergey Vlasov <vsu@altlinux.ru> :
In my testing with "qemu -net user" the 226 response to RETR was
often received earlier than final packets of the data connection;
this caused the received file to become truncated without any error
indication. Fix this by adding an intermediate state FTP_TRANSFER
between FTP_RETR and FTP_QUIT, so that the transfer is considered to
be complete only when both the end of data connection is encountered
and the final reply to the RETR command is received.
Verifying server ID and DHCP transaction ID is insufficient to
differentiate between DHCPACK and ProxyDHCPACK when the DHCP server and
Proxy DHCP server are the same machine.
dhcppkt_store() is supposed to clear the setting if passed NULL for the
setting data. In the case of fixed-location fields (e.g. client IP
address), this requires setting the content of the field to all-zeros.
Perform the same test for a matching DHCP_SERVER_IDENTIFIER on
ProxyDHCPACKs as we do for DHCPACKs. Otherwise, a retransmitted
DHCPACK can end up being treated as the ProxyDHCPACK.
I have a vague and unsettling memory that this test was deliberately
omitted, but I can't remember why, and can't find anything in the VC
logs.
Add the definition of SLAM_MAX_BLOCKS_PER_NACK, which is roughly
equivalent to a TCP window size; it represents the maximum number of
packets that will be requested in a single NACK.
Note that, to keep the code size down, we still limit ourselves to
requesting only a single range per NACK; if the missing-block list is
discontiguous then we may request fewer than SLAM_MAX_BLOCKS_PER_NACK
blocks.
On any fast network, or with any driver that may drop packets
(e.g. Infiniband, which has very small RX rings), the traditional
usage of the SLAM protocol will result in enormous numbers of packet
drops and a consequent large number of retransmissions.
By adapting the client behaviour, we can force the server to act more
like a multicast TFTP server, with flow control provided by a single
master client.
This behaviour should interoperate with any traditional SLAM client
(e.g. Etherboot 5.4) on the network. The SLAM protocol isn't actually
documented anywhere, so it's hard to define either behaviour as
compliant or otherwise.
A missing test for dhcp->dhcpoffer in dhcp_timer_expired() was causing
the client to transition to DHCPREQUEST after timing out on waiting
for ProxyDHCP even if no DHCPOFFERs had been received.
In a SLAM NACK packet, if we run out of space to represent the
missing-block list, then indicate all remaining blocks as missing.
This avoids the need to wait for the one-second timeout before
receiving the blocks that otherwise wouldn't have been requested due
to running out of space.
Shorter NACK packets take less time to construct and spew out less
debug output, and there's a limit to how useful it is to send a
complete missing-block list anyway; if the loss rate is high then
we're going to have to retransmit an updated missing-block list
anyway.
Also add pretty debugging output to show the list of requested blocks.
UDP sockets can be used for multicast, at which point it becomes
plausible that we could receive packets that aren't destined for us
but that still match on a port number.
Maintain state for the advertised window length, and only ever increase
it (instead of calculating it afresh on each transmit). This avoids
triggering "treason uncloaked" messages on Linux peers.
Respond to zero-length TCP keepalives (i.e. empty data packets
transmitted outside the window). Even if the peer wouldn't otherwise
expect an ACK (because its packet consumed no sequence space), force an
ACK if it was outside the window.
We don't yet generate TCP keepalives. It could be done, but it's unclear
what benefit this would have. (Linux, for example, doesn't start sending
keepalives until the connection has been idle for two hours.)
Return the most appropriate of EACCES, EPERM, ENODEV, ENOTSUP, EIO or
EINVAL depending on the exact error returned by the target, rather than
just always returning EPERM.
Also, ensure that error strings exist for these errors.
From: Viswanath Krishnamurthy <viswa.krish@gmail.com>
The current ipv4 incorrectly checks the IP address for multicast address.
This causes valid IPv4 unicast address to be trated as multicast address
For e.g if the PXE/tftp server IP address is 192.168.4.XXX where XXX is
224 or greater, it gets treated as multicast address and a ethernet
multicast address is sent out on the wire causing timeouts
Some EMC targets will fail if we advertise that we can authenticate with
CHAP, but the target is configured to allow unauthenticated access to that
target. We advertise AuthMethod=CHAP,None; the target should (I think)
select AuthMethod=None for unprotected targets. IETD does this, but an
EMC Celerra NS83 doesn't.
Fix by offering only AuthMethod=None if the user hasn't supplied a
username and password; this means that we won't be offering CHAP
authentication unless the user is expecting to use it (in which case the
target is presumably configured appropriately).
Many thanks to Alessandro Iurlano <alessandro.iurlano@gmail.com> for
reporting and helping to diagnose this problem.
Infiniband devices no longer block waiting for link-up in
register_ibdev().
Hermon driver needs to create an event queue and poll for link-up events.
Infiniband core needs to reread MAD parameters when link state changes.
IPoIB needs to cope with Infiniband link parameters being only partially
available at probe and open time.
gPXE is not compliant with the HTTP/1.1 specification (RFC 2616),
since it lacks support for "Transfer-Encoding: chunked". gPXE is,
however, compliant with the HTTP/1.0 specification (RFC 1945), which
does not require "Transfer-Encoding: chunked" to be supported.
The only HTTP/1.1 feature that gPXE uses is the "Host:" header, but
servers universally accept that one from HTTP/1.0 clients as an
optional extension (it is obligatory for HTTP/1.1). gPXE does not,
for example, appear to support connection caching. Advertising as a
HTTP/1.0 client will typically make the server close the connection
immediately upon sending the last data, which is actually beneficial
if we aren't going to keep the connection alive anyway.
The PXE spec is (as usual) unclear on precisely when ProxyDHCPREQUESTs
should be issued. We adapt the following, slightly paranoid approach:
If an offer contains an IP address, then it is a normal DHCPOFFER.
If an offer contains an option #60 "PXEClient", then it is a
ProxyDHCPOFFER. Note that the same packet can be both a normal
DHCPOFFER and a ProxyDHCPOFFER.
After receiving the normal DHCPACK, if we have received a
ProxyDHCPOFFER, we unicast a ProxyDHCPREQUEST back to the ProxyDHCP
server on port 4011. If we time out waiting for a ProxyDHCPACK, we
treat this as a non-fatal error.
Allow for settings to be described by something other than a DHCP option
tag if desirable. Currently used only for the MAC address setting.
Separate out fake DHCP packet creation code from dhcp.c to fakedhcp.c.
Remove notion of settings from dhcppkt.c.
Rationalise dhcp.c to use settings API only for final registration of the
DHCP options, rather than using {store,fetch}_setting throughout.
Add dedicated functions create_dhcpdiscover(), create_dhcpack() and
create_proxydhcpack() for use by external code such as the PXE preboot
code.
Register ProxyDHCP options under the global scope "proxydhcp".
Unregister previously-acquired DHCP and ProxyDHCP settings when DHCP
succeeds.
Add a configuration settings block for each net device. This will
provide the parent scope for settings applicable only to that network
device (e.g. non-volatile options stored on the NIC, options obtained via
DHCP, etc.).
Expose the MAC address as a setting.
RFC 4390 provides for the DHCP client identifier to contain the link-layer
hardware type and MAC address when the MAC address exceeds 16 bytes.
However, the hardware type field is only 8 bits; we were assuming 16 bits.
Arbel and Hermon cards both have multiple ports. Add the
infrastructure required to register each port as a separate IB
device. Don't yet register more than one port, since registration
will currently fail unless a valid link is detected.
Use ib_*_{set,get}_{drv,owner}data wrappers to access driver- and
owner-private data on Infiniband structures.
Pull out common code for handling management datagrams from arbel.c
and hermon.c into infiniband.c.
Add port number to struct ib_device.
Add open(), close() and mad() methods to struct ib_device_operations.
Allow port numbers in iSCSI redirection.
Wait for SCSI status, not just the final data-in (which may be followed
by an explicit SCSI Response PDU if the S bit is not set).
We didn't specify values for MaxRecvDataSegmentLength and
MaxBurstLength (to save space, since we were happy with the
RFC-defined default values of 8kB and 256kB respectively). However,
the OpenSolaris target (incorrectly) assumes default values of zero
for these parameters.
The upshot was that the OpenSolaris target would get stuck in an
endless loop trying to send us the first 512-byte sector, zero bytes
at a time, and would eventually run out of memory and core-dump.
Fixed by explicitly specifying the default values for these two
parameters.
tracking down a bug that turned out to be a free_iob() used where I
needed a netdev_tx_complete(). This left the freed I/O buffer on the
net device's TX list, with bad, bad consequences later.
Also fixed the bug in question.
safe dropping of the netdev ref by the driver while other refs still
exist.
Add netdev_irq() method. Net device open()/close() methods should no
longer enable or disable IRQs.
Remove rx_quota; it wasn't used anywhere and added too much complexity
to implementing correct interrupt-masking behaviour in pxe_undi.c.
request(), seek() or deliver_xxx() in order to start the data flow.
Autonomous generators must be genuinely autonomous (having their own
process), or otherwise arrange to be called. TCP does this by
starting the retry timer immediately.
Add some debugging statements.
will enable us to cascade async operations, which is necessary in order to
properly support DNS. (For example, an HTTP request may have to redirect
to a new location and will have to perform a new DNS lookup, so we can't
just rely on doing the name lookup at the time of parsing the initial
URL).
Anything other than HTTP is probably broken right now; I'll fix the others
up asap.
tcpip_tx(). This avoids the irritating wait when you accidentally type
"kernel pxelinux.0" before bringing up the network interface.
Add ENETUNREACH to strerror()'s list.
names.
Add "dev" pointer in struct net_device to tie network interfaces back to a
hardware device.
Force natural alignment of data types in __table() macros. This seems to
prevent gcc from taking the unilateral decision to occasionally increase
their alignment (which screws up the table packing).
"when SYN is ACKed and we have already received SYN", or
"when SYN is received and we have already had SYN ACKed"
rather than just
"when SYN is ACKed"
This avoids spuriously calling the connected() method when we receive
a RST,ACK in response to a SYN.
may be called during the RX data processor, before the RX engine
transitions back to idle. It doesn't really matter if the RX engine
isn't idle when iscsi_done() is called, because it will just pick up
where it left off on the next call. (The same is not true for the TX
engine, so keep the TX engine assertion.)
Truncate TX length to TCP window at time of transmission rather than at
time of adding to TX packet; this is conceptually cleaner and also allows
the application to call tcp_send() multiple times to build up a single
packet.
"flags that are currently being sent". This allows at least one special
case (checking that we haven't already sent a FIN in tcp_rx_fin()) to be
collapsed.
phase. In iscsi_issue(), detect this state and immediately refuse the
operation. This avoids trying multiple logins when scsi.c tries
several times to read the drive capacity.
defined in vsprintf.h. (This may change, since vsprintf.h is a
non-standard name, but for now it's the one to use.)
There should be no need to include vsprintf.h just for DBG() statements,
since include/compiler.h forces it in for a debug build anyway.
already stopped.
Don't call start_timer() when sending a dataless ACK. This may or may
not be the right thing to do; I can't tell.
Back out broken "send ACK only if required to" logic temporarily.
Added a large number of missing calls to free_pkb(). In the case of UDP,
no received packets were ever freed, which lead to memory exhaustion
remarkably quickly once pxelinux started up.
In general, any function with _rx() in its name which accepts a pk_buff
*must* either call free_pkb() or pass the pkb to another _rx() function
(e.g. the next layer up the stack). Since the UDP (and TCP) layers don't
pass packet buffers up to the higher-layer protocols (the
"applications"), they must free the packet buffer after calling the
application's newdata() method.
udp_connect() now follows the standard BSD sockets semantics and simply
sets the default address for outgoing packets; it doesn't filter incoming
packets.
Updated PXE API dispatcher to use copy_{to,from}_user, and moved to
arch/i386 since the implementation is quite architecture-dependent.
(The individual PXE API calls can be largely
architecture-independent.)
Parameter request list is yet another option that ISC dhcpd ignores
unless it's in the main options block.
Fix logic error in parsing file and sname fields.
Use pkb_available() rather than pkb_len() (which will always return 0
on a freshly allocated buffer).
udp_send() should assume that the buffer has already been allocated.
structure, and populating it with options. This should allow us to
use the same basic options list for both DHCPDISCOVER and DHCPREQUEST,
plus making it much easier to set the non-constant parameters
(e.g. requested IP address) in request packets.
to search through all registered option blocks.
Move some static inlines that are likely to be used frequently into
dhcpopts.c as normal functions, to save space.
is a DHCP option" data structures.
We need this code in order to be able to return a DHCP packet to a PXE NBP
which reflects options from our multiple sources (e.g. NVS and DHCP
server). This is expensive, but necessary. Having paid this cost, we may
as well try to use the same code to generate our DHCP request packets,
since the process is similar.
Kill off the static single net device and move to proper dynamic
registration (which we need with the new device model).
Break the (flawed) assumption that all network-layer protocols can use
ARP; such network-layer protocols (i.e. IPv4) must now register as an ARP
protocol using ARP_NET_PROTOCOL() and provide a single method for checking
the existence of a local network-layer address.
async_wait(), though we may wish to move this higher up the stack, and
consider making the block device model asynchronous. (There is only a
marginal cost for synchronous devices, since they can simply call
async_done() before returning; async_wait() will work seamlessly in this
situation).
Removed data_in_len and data_out_len from ata_command structure; the
lengths are implied by the sector count and the presence of the data_in
or data_out pointers.
Changed AoE code to use subcommands by default, and made aoe_issue()
nonblocking (with completion via async_wait()).
these timer objects in AoE and UDP protocols (where there is no underlying
retransmission mechanism) without requiring each protocol to implement its
own individual retry logic. Eventually, we should be able to use the same
timer code for TCP retransmissions as well.
Cannot yet handle reads of more than two sectors
No retransmission
No way to find out a target's MAC address (this proof of concept uses
broadcasts)
These limitations shall not last long! :)
the kernel), which encapsulates the information needed to refer to an
external buffer. Under normal operation, this can just be a void *
equivalent, but under -DKEEP_IT_REAL it would be a segoff_t equivalent.
Use this concept to avoid the need for bounce buffers in int13.c,
which reduces memory usage and opens up the possibility of using
multi-sector reads.
Extend the block-device API and the SCSI block device implementation
to support multi-sector reads.
Update iscsi.c to use user buffers.
Move the obsolete portions of realmode.h to old_realmode.h.
MS-DOS now boots an order of magnitude faster over iSCSI (~10 seconds
from power-up to C:> prompt in bochs).
errors; still need to verify data integrity).
SCSI response PDUs are handled: status and sense data (if available) are
returned via the scsi_command structure.
Updated iSCSI session parameter usage.
take ownership of the packet, rather than doing so only if they return
success. This breaks semantic compatibility with Linux's
hard_start_xmit() method, but means that we don't have to worry so much
about error cases.
Split mechanism of processing received packets (net_rx_process()) out
from policy (net_step()), preparatory to putting net_step() in a separate
object.
implementation allows for only one, and does so without compromising on
the efficiency of static allocation).
Link-layer protocols are cleanly separated from the device drivers.
Network-layer protocols are cleanly separated from individual network
devices.
Link-layer and network-layer protocols are cleanly separated from each
other.
implementation allows for only one, and does so without compromising on
the efficiency of static allocation).
Link-layer protocols are cleanly separated from the device drivers.
Network-layer protocols are cleanly separated from individual network
devices.
Link-layer and network-layer protocols are cleanly separated from each
other.
Michael Brown
Joshua Oreman
Daniel Verkamp
Stefan Hajnoczi
Andrew Schran
Sergey Vlasov
H. Peter Anvin
Marty Connor
Alexey Zaytsev
Holger Lubitz
Anselm Martin Hoffmeister
Nikhil Chandru Rao
Derek Pryor