Michael Brown [Sat, 26 Mar 2011 14:58:17 +0000 (14:58 +0000)]
[tcp] Record ts_recent on first received packet
Commit 6861304 ("[tcp] Handle out-of-order received packets")
introduced a regression in which ts_recent would not be updated until
the first packet is received in the ESTABLISHED state, i.e. the
timestamp from the SYN+ACK packet would be ignored. This causes the
connection to be dropped by strictly-conforming TCP peers, such as
FreeBSD.
Fix by delaying the timestamp window check until after processing the
received SYN flag.
Reported-by: winders@sonnet.com Signed-off-by: Michael Brown <mcb30@ipxe.org>
Michael Brown [Wed, 23 Mar 2011 13:51:32 +0000 (13:51 +0000)]
[undi] Cope with devices that erroneously claim not to use interrupts
Some PXE stacks advertise that interrupts are not supported, despite
requiring the use of interrupts. Attempt to cope with such cards
without breaking others by always hooking the interrupt, and using the
"interrupts supported" flag only to decide whether or not to wait for
an interrupt before calling PXENV_UNDI_ISR_IN_PROCESS.
The possible combinations are therefore:
1. Card generates interrupts and claims to support interrupts
iPXE will call PXENV_UNDI_ISR_IN_PROCESS only after an interrupt
has been observed. (This is required to avoid lockups in some PXE
stacks, which spuriously sulk if called before an interrupt has
been generated.)
Such a card should work correctly.
2. Card does not generate interrupts and does not claim to support
interrupts
iPXE will call PXENV_UNDI_ISR_IN_PROCESS indiscriminately, matching
the observed behaviour of at least one other PXE NBP (winBoot/i).
Such a card should work correctly.
3. Card generates interrupts but claims not to support interrupts
iPXE will call PXENV_UNDI_ISR_IN_PROCESS indiscriminately. An
interrupt will still result in a call to PXENV_UNDI_ISR_IN_START.
Such a card may work correctly.
4. Card does not generate interrupts but claims to support interrupts
Such a card will not work at all.
Reported-by: Jerry Cheng <jaspers.cheng@msa.hinet.net> Tested-by: Jerry Cheng <jaspers.cheng@msa.hinet.net> Reported-by: Mauricio Silveira <mauricio@livreti.com.br> Tested-by: Mauricio Silveira <mauricio@livreti.com.br> Signed-off-by: Michael Brown <mcb30@ipxe.org>
Michael Brown [Wed, 23 Mar 2011 10:08:16 +0000 (10:08 +0000)]
[settings] Fix setting expansion
Commit 5fbd020 ("[settings] Display canonical setting name in output
of "show" command") introduced a regression causing all setting
expansions (e.g. "${net0/mac}") to expand to an empty string.
Fix by returning the formatted value length from
fetchf_named_setting(), as expected by the caller.
Michael Brown [Tue, 22 Mar 2011 21:07:22 +0000 (21:07 +0000)]
[settings] Display only applicable settings in "config" user interface
Display only settings relevant to the current scope. For example,
"config net0" no longer displays SMBIOS settings, and "config smbios"
displays only SMBIOS settings.
Originally-implemented-by: Glenn Brown <glenn@myri.com> Signed-off-by: Michael Brown <mcb30@ipxe.org>
Michael Brown [Wed, 16 Mar 2011 21:31:16 +0000 (21:31 +0000)]
[console] Try to avoid problems caused by keycode 86
The "us" keyboard layout contains a mapping for keycode 86 (which
seems not to correspond to any physical key on many US keyboards) to
the ASCII character '<'. This mapping causes conflicts with the
mapping for keycode 51, which also maps (with shift) to '<'.
Change the keyboard mapping generator to choose the lowest keycode for
each ASCII character as indicating the relevant mapping to use, on the
basis that a lower keycode roughly indicates a "more normal" key. On
a German keyboard, which has keys for both keycode 51 and keycode 86
present, this causes '<' to be remapped to ';', which is a closer
match to typical user expectations.
Reported-by: Sven Dreyer <sven@dreyer-net.de> Signed-off-by: Michael Brown <mcb30@ipxe.org>
Michael Brown [Wed, 16 Mar 2011 19:51:02 +0000 (19:51 +0000)]
[eepro] Avoid unused variable warning in gcc 4.6
From a cursory examination, it appears as though the calculation of
tx_available is redundant, since eepro_transmit() waits for transmit
completion before returning anyway.
Reported-by: Ralph Giles <giles@thaumas.net> Tested-by: Ralph Giles <giles@thaumas.net> Signed-off-by: Michael Brown <mcb30@ipxe.org>
Michael Brown [Wed, 16 Mar 2011 18:58:52 +0000 (18:58 +0000)]
[e1000] Remove unused variable when ICR register is simply cleared
On reset and close, the ICR register is read to clear any pending
interrupts, but the value is simply ignored. Avoid assigning the
value to a variable, to inhibit a warning from gcc 4.6.
Also fix a potential race condition in reset routines which clear
interrupts before disabling them.
Reported-by: Ralph Giles <giles@thaumas.net> Tested-by: Ralph Giles <giles@thaumas.net> Signed-off-by: Michael Brown <mcb30@ipxe.org>
Michael Brown [Wed, 16 Mar 2011 18:10:49 +0000 (18:10 +0000)]
[build] Avoid spurious address comparison warnings in gcc 4.6
A construction such as "assert ( ptr != NULL )" seems to trigger a
false positive warning in gcc 4.6 if the value of "ptr" is known at
compile-time to be non-NULL. Use -Wno-address to inhibit this
warning.
Reported-by: Ralph Giles <giles@thaumas.net> Tested-by: Ralph Giles <giles@thaumas.net> Signed-off-by: Michael Brown <mcb30@ipxe.org>
Michael Brown [Wed, 16 Mar 2011 17:15:25 +0000 (17:15 +0000)]
[build] Avoid spurious unused-but-set-variable warnings in gcc 4.6
The __table_entries() construction seems to trigger a false positive
warning in gcc 4.6 relating to variables which are set but never
used. Add __attribute__((unused)) to inhibit this warning.
Reported-by: Ralph Giles <giles@thaumas.net> Tested-by: Ralph Giles <giles@thaumas.net> Signed-off-by: Michael Brown <mcb30@ipxe.org>
The keymap files, though autogenerated, are checked in to version
control and should be considered as source files. They should never
be automatically rebuilt.
Michael Brown [Tue, 15 Mar 2011 16:55:04 +0000 (16:55 +0000)]
[efi] Mark SNP devices as children of EFI PCI device
Re-open the EFI_PCI_IO_PROTOCOL specifying an Attributes value of
EFI_OPEN_PROTOCOL_BY_CHILD_CONTROLLER. This causes the SNP devices to
be marked as children of the EFI PCI device (as shown in the "devtree"
command).
On at least one IBM blade system, this is required in order to have
the relevant drivers automatically attach to the SNP controller at
device creation time.
Michael Brown [Tue, 15 Mar 2011 01:52:58 +0000 (01:52 +0000)]
[romprefix] Use correct size for PMM allocation of image source block
iPXE allocates its first PMM block using the image source length,
which is rounded up to the nearest 16-byte paragraph. It then copies
in data of a length calculated from the ROM size, which is
theoretically less than or equal to the image source length, but is
rounded up to the nearest 512-byte sector. This can result in copying
beyond the end of the allocated PMM block, which can corrupt the PMM
data structures (and other essentially arbitrary areas of memory).
Fix by rounding up the image source length to the nearest 512-byte
sector before using it as the PMM allocation length.
Reported-by: Alex Williamson <alex.williamson@redhat.com> Reported-by: Jarrod Johnson <jarrod.b.johnson@gmail.com> Reported-by: Itay Gazit <itayg@mellanox.co.il> Signed-off-by: Michael Brown <mcb30@ipxe.org>
Michael Brown [Sat, 12 Mar 2011 12:48:29 +0000 (12:48 +0000)]
[eepro100] Do not discard short packets
In a virtual environment such as qemu, we can legitimately receive
packets less than 64 bytes in length, such as ARP replies. These are
currently discarded, causing most IPv4 communication to fail.
Fix by ignoring the RFDShort bit when receiving packets.
Reported-by: Alex Williamson <alex.williamson@redhat.com> Signed-off-by: Michael Brown <mcb30@ipxe.org>
Michael Brown [Thu, 10 Mar 2011 22:50:05 +0000 (22:50 +0000)]
[bios] Use INT 16,11 instead of INT 16,01 to check for keystrokes
INT 16,01 will discard some extended keystrokes on some BIOSes, making
it impossible for iPXE to detect keypresses such as F12. Fix by using
INT 16,11 instead.
Michael Brown [Wed, 9 Mar 2011 17:32:32 +0000 (17:32 +0000)]
[script] Allow scripts to be replaced
A script that downloads a new image using imgdownload() with the
action register_and_replace_image() can now be freed immediately
before the replacement image is executed. This functionality is not
yet exposed via an iPXE command.
Michael Brown [Mon, 7 Mar 2011 22:02:07 +0000 (22:02 +0000)]
[prefix] Allow iPXE's own command line to be executed as a script
Some prefixes (e.g. .lkrn) allow a command line to be passed in to
iPXE. At present, this command line is ignored.
If a command line is provided, treat it as an embedded script (without
an explicit "#!ipxe" magic marker). This allows for patterns of
invocation such as
title iPXE
kernel /boot/ipxe.lkrn dhcp && \
sanboot iscsi:10.0.4.1::::iqn.2010-04.org.ipxe.dolphin:storage
Here GRUB is instructed to load ipxe.lkrn with an embedded script
equivalent to
Michael Brown [Mon, 7 Mar 2011 18:32:30 +0000 (18:32 +0000)]
[script] Add "prompt" command
The "prompt" command exposes the prompt() function, allowing a script
to prompt the user for a keypress and take action depending on the
result. For example
#!ipxe
prompt -k 0x197e -t 2000 Press F12 to boot from network... || exit
autoboot
Michael Brown [Mon, 7 Mar 2011 19:17:51 +0000 (19:17 +0000)]
[bios] Recognise scancodes for F5-F12 inclusive
The function keys F5-F12 all conform to the same ANSI pattern as the
other "special" keys that we currently recognise. Add these key
definitions, and shrink the representation of the ANSI sequences in
bios_console.c to compensate.
Michael Brown [Mon, 7 Mar 2011 00:37:50 +0000 (00:37 +0000)]
[image] Simplify image management
Refactor the {load,exec} image operations as {probe,exec}. This makes
the probe mechanism cleaner, eliminates some forward declarations,
avoids holding magic state in image->priv, eliminates the possibility
of screwing up between the "load" and "exec" stages, and makes the
documentation simpler since the concept of "loading" (as distinct from
"executing") no longer needs to be explained.
Michael Brown [Fri, 4 Mar 2011 19:17:53 +0000 (19:17 +0000)]
[spi] Reset device on each access
When chainloading rtl8139.pxe from an old Etherboot rtl8139.zrom, iPXE
can end up misreading the first word of the MAC address from the
EEPROM as being all zeroes. This is presumably because Etherboot has
left the serial EEPROM in an unexpected state.
Fix by using the chip select line to reset the SPI device before we
start accessing it.
Reported-by: Mandar U Jog <mandarjog@gmail.com> Tested-by: Mandar U Jog <mandarjog@gmail.com> Signed-off-by: Michael Brown <mcb30@ipxe.org>
Michael Brown [Fri, 4 Mar 2011 12:14:51 +0000 (12:14 +0000)]
[parseopt] Refer to online documentation for command help
The online documentation (e.g. http://ipxe.org/cmd/ifopen), though not
yet complete, is far more comprehensive than could be provided within
the iPXE binary. Save around 200 bytes (compressed) by removing the
command descriptions from the interactive help, and instead referring
users directly to the web page describing the relevant command.
Michael Brown [Thu, 3 Mar 2011 22:08:50 +0000 (22:08 +0000)]
[iscsi] Change default initiator IQN
The default initiator IQN is "iqn.2000-09.org.etherboot:UNKNOWN".
This is problematic for two reasons:
a) the etherboot.org domain (and hence the associated IQN namespace)
is not under the control of the iPXE project, and
b) some targets (correctly) refuse to allow concurrent connections
from different initiators using the same initiator IQN.
Solve both problems by changing the default initiator IQN to be
iqn.2010-04.org.ipxe:<hostname> if a hostname is set, or
iqn.2010-04.org.ipxe:<uuid> if no hostname is set.
Explicit initiator IQNs set via DHCP option 203 are not affected by
this change.
Unfortunately, this change is likely to break some existing
configurations, where ACL rules have been put in place referring to
the old default initiator IQN. Users may need to update ACLs, or
force the use of the old IQN using an iPXE script line such as
set initiator-iqn iqn.2000-09.org.etherboot:UNKNOWN
Michael Brown [Thu, 3 Mar 2011 19:59:31 +0000 (19:59 +0000)]
[settings] Make fetch_string_setting_copy() easier to use
Most callers of functions in the fetch_setting() family treat any
errors as meaning "non-existent setting". In the case of
fetch_string_setting_copy(), an existent setting can still result in
an error due to memory allocation failure.
Allow the caller to distinguish between a non-existent setting and an
error in allocating memory for the copy, by returning success (and a
NULL buffer pointer) for a non-existent setting.
Michael Brown [Thu, 3 Mar 2011 17:13:44 +0000 (17:13 +0000)]
[settings] Avoid fetch_string_setting_copy() leaving an uninitialised pointer
For consistency with other functions in the fetch_setting() family,
ensure that fetch_string_setting_copy() always initialises the pointer
to the fetched setting even if fetching fails.
Michael Brown [Wed, 23 Feb 2011 17:52:13 +0000 (17:52 +0000)]
[bofm] Allow garbage collection of BOFM code in non-BOFM builds
Most builds will not have BOFM enabled. In these builds, allow all
BOFM code (including BOFM-only code within the individual drivers) to
be garbage-collected at link time in order to save space in the final
binary.
Michael Brown [Thu, 10 Feb 2011 21:34:34 +0000 (21:34 +0000)]
[bofm] Add offline BOFM test facility
Testing BOFM involves gaining access to an IBM blade chassis, which is
often not practical. Provide a facility for testing BOFM
functionality outside of a real IBM blade context.
Michael Brown [Wed, 2 Mar 2011 19:57:01 +0000 (19:57 +0000)]
[autoboot] Allow a SAN boot as a fallback if a filename boot returns
Currently, if both a filename and root-path are present, iPXE will
hook the SAN device but will only attempt to boot from the filename.
Change this behaviour so that both are attempted. Users who want to
avoid booting from the SAN as a fallback can do so via the existing
"skip-san-boot" setting.
This allows for seamless deployment to a SAN target using Windows
Deployment Services (and similar products). A user simply has to
define the root-path option in DHCP and then use WDS to deploy the
system. No further configuration should be required.
Michael Brown [Wed, 2 Mar 2011 19:39:39 +0000 (19:39 +0000)]
[autoboot] Tidy up output following NBP execution
If the NBP returns, then always print a trailing newline, since some
NBPs (e.g. wdsnbp.com) leave the cursor in a random position halfway
across the screen.
Michael Brown [Wed, 2 Mar 2011 19:29:24 +0000 (19:29 +0000)]
[image] Allow download job to complete before acting upon image
Allow the monojob controlling the download to complete before calling
register_image() and friends. This allows the trailing "ok" from
monojob.c to be printed before the image starts executing (and
possibly printing output of its own).
Michael Brown [Wed, 2 Mar 2011 15:33:39 +0000 (15:33 +0000)]
[int13] Automatically reopen underlying block device as needed
We currently use INT 13,00 as an opportunity to reopen the underlying
block device, which works well for callers such as DOS that will use
INT 13,00 in response to any disk errors. However, some callers (such
as Windows Server 2008) do not attempt to reset the disk, and so any
failures become effectively permanent.
Fix this by automatically reopening the underlying block device
whenever we might want to access it.
This makes direct installation of Windows to an iSCSI target much more
reliable.
Michael Brown [Tue, 1 Mar 2011 16:03:44 +0000 (16:03 +0000)]
[prefix] Set the "size" bit in the GDT entry for the flat data segment
The "size" bit (aka the D/B) bit should (as far as I can tell) be
irrelevant for accesses to a non-code, non-stack, expand-upwards
segment. However, VirtualBox fails on some accesses via this segment
if this bit is not set.
This change allows iPXE to boot under VirtualBox without having to
disable VT-x/AMD-V support.
projects / thirdparty / ipxe.git / log
