heap when using a range above 1MB.
Previously the loader would always use the last 3MB in the first memory
range above 1MB for the heap. However, this memory range is also where the
kernel and any modules are loaded. If this memory range is "small", then
using the high 3MB for the heap may not leave enough room for the kernel
and modules.
Now the loader will use any range below 4GB for the heap, and the logic to
choose the "high" heap region has moved into biosmem.c. It sets two
variables that the loader can use for a high heap if it desires. When a
high heap is enabled (BZIP2, FireWire, GPT, or ZFS), then the following
memory ranges are preferred for the heap in order from best to worst:
- The largest memory region in the SMAP with a start address greater than
1MB. The memory region must be at least 3MB in length. This leaves the
region starting at 1MB purely for use by the kernel and modules.
- The last 3MB of the memory region starting at 1MB if it is at least 3MB
in size. This matches the current behavior except that the current loader
would break horribly if the first region was not at least 3MB in size.
- The memory range from the end of the loader up to the 640k window. This
is the range the loader uses when none of the high-heap-requesting options
are enabled.
Tested by: hrs
MFC after: 1 week
video console which doesn't take any input from keyboard and hides
all output replacing it with ``spinning'' character (useful for
embedded products and custom installations).
Sponsored by: Sippy Software, Inc.
This adds zfsloader which will be called by zfsboot/gptzfsboot code rather
than the tradional loader. This eliminates the need to set the
LOADER_ZFS_SUPPORT variable in order to get a ZFS enabled loader.
Note however, that you must reinstall your bootcode (zfsboot/gptzfsboot)
in order for the boot process to use the new loader.
New installations will no longer be required to build a ZFS enabled
loader for a working ZFS boot system. Installing zfsboot/gptzfsboot is
sufficient for acknowledging the use of CDDL code and therefore the ZFS
enabled loader.
Based on a previous patch from jhb@
Reviewed by: jhb@
MFC after: 2 weeks
fully support booting from large volumes.
Tested by: Emil Smolenski ambsd of raisa.eu.org
Submitted by: Matt Reimer mattjreimer of gmail (most of the C bits)
MFC after: 1 week
only when typing the sequence "123" (opposite to the standard 'push any
button' approach).
That results useful when using serial lines sending garbage and leading
to unwilling boot prompt appearence.
Obtained from: Sandvine Incorporated
Reviewed by: emaste, jhb
Sponsored by: Sandvine Incorporated
MFC: 1 week
- Teach it to read gang blocks. (essentially untested)
If you see "ZFS: gang block detected!", please let
me know, so we can either remove the printf if it
works, or fix it if it doesn't.
- If multiple partitions exist on a disk, probe them all.
We also need to reset dsk->start to 0 to read the right
sector here.
- With GPT, we can have 128 partitions.
- If the bootfs property has ever been set on a pool
it seems that it never goes away. zpool won't allow
you to add to the pool with the bootfs property set.
However, if you clear the property back to default
we end up getting 0 for the object number and read
a bogus block pointer and fail to boot.
- Fix some error printfs. The printf in the loader is
only capable of c,s and u formats.
- Teach printf how to display %llu
Reviewed by: dfr, jhb
MFC after: 2 weeks
short read requests, so the result was that a /boot.config smaller than 512
bytes was ignored. boot2 uses fsread() instead of xfsread() to read
/boot.config already, so this makes zfsboot more like boot2.
Submitted by: Johny Mattsson johny-freebsd of earthmagic org
Reviewed by: dfr
MFC after: 3 days
devices that we also support, just not by default (thus only LINT or
module builds by default).
While currently there is only "/dev/full" [2], we are planning to see more
in the future. We may decide to change the module/dependency logic in the
future should the list grow too long.
This is not part of linux.ko as also non-linux binaries like kFreeBSD
userland or ports can make use of this as well.
Suggested by: rwatson [1] (name)
Submitted by: ed [2]
Discussed with: markm, ed, rwatson, kib (weeks ago)
Reviewed by: rwatson, brueffer (prev. version)
PR: kern/68961
MFC after: 6 weeks
things a bit:
- use dpcpu data to track the ifps with packets queued up,
- per-cpu locking and driver flags
- along with .nh_drainedcpu and NETISR_POLICY_CPU.
- Put the mbufs in flight reference count, preventing interfaces
from going away, under INVARIANTS as this is a general problem
of the stack and should be solved in if.c/netisr but still good
to verify the internal queuing logic.
- Permit changing the MTU to virtually everythinkg like we do for loopback.
Hook epair(4) up to the build.
Approved by: re (kib)
slicei, Apple EFI hardware), the bootloader will fail to recognize the GPT
if it finds anything else but the EFI partition. Change the check to continue
detecting the GPT by looking at the EFI partition on the MBR but
stopping successfuly after finding it.
PR: kern/134590
Submitted by: Christoph Langguth <christoph at rosenkeller.org>
Reviewed by: jhb
MFC after: 2 weeks
Approved by: re (kib)
DP83065 Saturn Gigabit Ethernet controllers. These are the successors
of the Sun GEM controllers and still have a similar but extended transmit
logic. As such this driver is based on gem(4).
Thanks to marcel@ for providing a Sun Quad GigaSwift Ethernet UTP (QGE)
card which was vital for getting this driver to work on architectures
not using Open Firmware.
Approved by: re (kib)
MFC after: 2 weeks
controller. These controllers are also known as L1C(AR8131) and
L2C(AR8132) respectively. These controllers resembles the first
generation controller L1 but usage of different descriptor format
and new register mappings over L1 register space requires a new
driver. There are a couple of registers I still don't understand
but the driver seems to have no critical issues for performance and
stability. Currently alc(4) supports the following hardware
features.
o MSI
o TCP Segmentation offload
o Hardware VLAN tag insertion/stripping
o Tx/Rx interrupt moderation
o Hardware statistics counters(dev.alc.%d.stats)
o Jumbo frame
o WOL
AR8131/AR8132 also supports Tx checksum offloading but I disabled
it due to stability issues. I'm not sure this comes from broken
sample boards or hardware bugs. If you know your controller works
without problems you can still enable it. The controller has a
silicon bug for Rx checksum offloading, so the feature was not
implemented.
I'd like to say big thanks to Atheros. Atheros kindly sent sample
boards to me and answered several questions I had.
HW donated by: Atheros Communications, Inc.
=================
Extend the loader to parse the root file system mount options in /etc/fstab,
and set a new loader variable vfs.root.mountfrom.options with these options.
The root mount options must be a comma-delimited string, as specified in
/etc/fstab.
Only set the vfs.root.mountfrom.options variable if it has not been
set in the environment.
sys/kern/vfs_mount.c
====================
When mounting the root file system, pass the mount options
specified in vfs.root.mountfrom.options, but filter out "rw" and "noro",
since the initial mount of the root file system must be done as "ro".
While we are here, try to add a few hints to the mountroot prompt
to give users and idea what might of gone wrong during mounting
of the root file system.
Reviewed by: jhb (an earlier patch)
uses the generic struct dirent, which happens to look identical to UFS's
struct direct. If BSD ever changes dirent then this will be a problem.
Submitted by: matthew dot fleming at isilon dot com
- Do not iterate int 15h, function e820h twice. Instead, we use STAILQ to
store each return buffer and copy all at once.
- Export optional extended attributes defined in ACPI 3.0 as separate
metadata. Currently, there are only two bits defined in the specification.
For example, if the descriptor has extended attributes and it is not
enabled, it has to be ignored by OS. We may implement it in the kernel
later if it is necessary and proven correct in reality.
- Check return buffer size strictly as suggested in ACPI 3.0.
Reviewed by: jhb
open partition. This fixes access to partitions whose starting offset
is >= 2 TB.
Submitted by: "James R. Van Artsdalen" james jrv.org
MFC after: 3 days
- First three fields of system UUID may be little-endian as described in
SMBIOS Specification v2.6. For now, we keep the network byte order for
backward compatibility (and consistency with popular dmidecode tool)
if SMBIOS table revision is less than 2.6. However, little-endian format
can be forced by defining BOOT_LITTLE_ENDIAN_UUID from make.conf(5) if it
is necessary.
- Replace overly ambitious optimizations with more readable code.
- Update comments to SMBIOS Specification v2.6 and clean up style(9) bugs.
as 'real memory' instead of Maxmem if the value is available.
Note amd64 displayed physmem as 'usable memory' since machdep.c r1.640
to unconfuse users. Now it is consistent across amd64 and i386 again.
While I am here, clean up smbios.c a bit and update copyright date.
Reviewed by: jhb
driver in Linux 2.6. uscanner was just a simple wrapper around a fifo and
contained no logic, the default interface is now libusb (supported by sane).
Reviewed by: HPS
booting because the CD driver did not use bounce buffers to ensure
request buffers sent to the BIOS were always in the first 1MB. Copy over
the bounce buffer logic from the BIOS disk driver (minus the 64k boundary
code for floppies) to fix this.
Reported by: kensmith
in make.conf or src.conf.
- When GPT is enabled (which it is by default), use memory above 1 MB and
leave the memory from the end of the bss to the end of the 640k window
purely for the stack. The loader has grown and now it is much more
common for the heap and stack to grow into each other when both are
located in the 640k window.
PR: kern/129526
MFC after: 1 week
the disklabel in the 2nd sector for boot code. Even with both UFS1
and UFS2 supported, there's enough bytes left that we don't have to
nibble from the disklabel.
Thus, the entire 2nd sector is now reserved for the disklabel, which
makes the bootcode compatible again with disklabels that have more
than 8 partitions -- such as those created and supported by gpart.
i386: 135 bytes available
amd64: 151 bytes available
Ok'd by: jhb
The old BTX passed the general purpose registers from the 32-bit client to
the routines called via virtual 86 mode. The new BTX did the same thing.
However, it turns out that some instructions behave differently in virtual 86
mode and real mode (even though this is under-documented). For example, the
LEAVE instruction will cause an exception in real mode if any of the upper
16-bits of %ebp are non-zero after it executes. In virtual 8086 mode the
upper 16-bits are simply ignored. This could cause faults in hardware
interrupt handlers that inherited an %ebp larger than 0xffff from the 32-bit
client (loader, boot2, etc.) while running in real mode.
To fix, when executing hardware interrupt handlers provide an explicit clean
state where all the general purpose and segment registers are zero upon
entry to the interrupt handler. While here, I attempted to simplify the
control flow in the 'intusr' code that sets up the various stack frames
and exits protected mode to invoke the requested routine via real mode.
A huge thanks to Tor Egge (tegge@) for debugging this issue.
Submitted by: tegge
Reviewed by: tegge
Tested by: bz
MFC after: 1 week
kernel one as the non-faulting flush address in the loader so
we can can change KERNBASE and VM_MIN_KERNEL_ADDRESS if we
ever want to without needing to worry about using a compatible
loader.
- Correctly check for LOADER_DEBUG.
- Add a missing const for page_sizes[].
functions used in the bootloader. The goal is to make the code more
readable and smaller (especially because we have size issues
in the loader's environment).
High level description of the changes:
+ define some string manipulation functions to improve readability;
+ create functions to manipulate module descriptors, removing some
duplicated code;
+ rename the error codes to ESOMETHING;
+ consistently use set_environment_variable (which evaluates
$variables) when interpreting variable=value assignments;
I have tested the code, but there might be code paths that I have
not traversed so please let me know of any issues.
Details of this change:
--- loader.4th ---
+ add some module operators, to remove duplicated code while parsing
module-related commands:
set-module-flag
enable-module
disable-module
toggle-module
show-module
--- pnp.4th ---
+ move here the definition related to the pnp devices list, e.g.
STAILQ_* , pnpident, pnpinfo
--- support.4th ---
+ rename error codes to capital e.g. ENOMEM EFREE ... and do obvious
changes related to the renaming;
+ remove unused structures (those relevant to pnp are moved to pnp.4th)
+ various string functions
- strlen removed (it is an internal function)
- strchr, defined as the C function
- strtype -- type a string to output
- strref -- assign a reference to the string on the stack
- unquote -- remove quotes from a string
+ remove reset_line_buffer
+ move up the 'set_environment_variable' function (which now
uses the interpreter, so $variables are evaluated).
Use the function in various places
+ add a 'test_file function' for debugging purposes
MFC after: 4 weeks
and re-enable it as default.
In particular:
+ re-enable the 'update' flag in the Makefile (of course!);
+ commit Warner's patch "orb $NOUPDATE,_FLAGS(%bp)"
to avoid writing to disk in case of a timeout/default choice;
+ fix an off-by-one count in the partition scan code that would
print the wrong name for unknown partitions;
+ unconditionally change the boot prompt to 'Boot:' instead of 'Default:'
to make room for the extra code/checks/messages. Some of the changes
listed below are also made to save space;
+ rearrange and fix comments for known partition types. Right now we
explicitly recognise *BSD, Linux, FAT16 (type 6, used on many USB keys),
NTFS (type 7), FAT32 (type 11).
Depending on other options we also recognise Extended (type 5),
FAT12 (type 1) and FAT16 < 32MB (type 4).
+ Add an entry "F6 PXE" when the code is built with -DPXE (which is
a default now). Technically, F6 boots through INT18, so the prompt 'PXE'
is a bit misleading. Unfortunately the name INT18
is too long and does not fit in - we could use ROM perhaps.
The reason I picked 'PXE' is that on many (I believe) new systems
INT18 calls PXE.
Apart from the choice of the name for PXE/ROM/INT18, this should close
pending issues on the 1-sector boot0 code and we should be able to
move the code to RELENG_7 when it reopens.
No boot0cfg changes are necessary.
MFC after: 3 weeks
loader_conf_files="foo bar baz"
should cause loading the files listed, and then resume with the
remaining config files (from previous values of the variable).
Unfortunately, sometimes the line was ignored -- actually even
modifying the line in /boot/default/loader.conf sometimes doesn't work.
ANALYSIS: After much investigation, turned out to be a bug in the logic.
The existing code detected a new assignment by looking at the address
of the the variable containing the string. This only worked by pure
chance, i.e. if the new string is longer than the previous value
then the memory allocator may return a different address
to store the string hence triggering the detection.
SOLUTION: This commit contains a minimal change to fix the problem,
without altering too much the existing structure of the code.
However, as a step towards improving the quality and reliability of
this code, I have introduced a handful of one-line functions
(strget, strset, strfree, string= ) that could be used in dozens
of places in the existing code.
HOWEVER:
There is a much bigger problem here. Even though I am no Forth
expert (as most fellow src committers) I can tell that much of the
forth code (in support.4th at least) is in severe need of a
review/refactoring:
+ pieces of code are replicated multiple times instead of writing
functions (see e.g. set_module_*);
+ a lot of stale code (e.g. "structure" definitions for
preloaded_files, kernel_module, pnp stuff) which is not used
or at least belongs elsewhere.
The code bload is extremely bad as the loader runs with very small
memory constraints, and we already hit the limit once (see
http://svn.freebsd.org/viewvc/base?view=revision&revision=185132
Reducing the footprint of the forth files is critical.
+ two different styles of coding, one using pure stack functions
(maybe beautiful but surely highly unreadable), one using
high level mechanisms to give names to arguments and local
variables (which leads to readable code).
Note that this code is used by default by all FreeBSD installations,
so the fragility and the code bloat are extremely damaging.
I will try to work fixing the three items above, but if others have
time, please have a look at these issues.
MFC after: 4 weeks
boot0.S changes:
+ import a patch from Christoph Mallon to rearrange the various
print functions and save another couple of bytes;
+ implement the suggestion in PR 70531 to enable booting from
any valid partition because even the extended partitions that
were previously in our kill list may contain a valid boot loader.
This simplifies the code and saves some bytes;
+ followwing up PR 127764, implement conditional code to preserve
the 'Volume ID' which might be used by other OS (NT, XP, Vista)
and is located at offset 0x1b8. This requires a relocation of the
parameter block within the boot sector -- there is no other
possible workaround.
To address this, boot0cfg has been updated to handle both
versions of the boot code;
+ slightly rearrange the strings printed in the menus to make
the code buildable with all options. Given the tight memory
budget, this means that with certain options we need to
shrink or remove certain labels.
and especially:
make -DVOLUME_LABEL -DPXE the default options.
This means that the newly built boot0 block will preserve the
Volume ID, and has the (hidden) option F6 to boot from INT18/PXE.
I think the extra functionality is well worth the change.
The most visible difference here is that the 'Default: ' string
now becomes 'Boot: ' (it can be reverted to the old value
but then we need to nuke 1/2 partition name or entries to
make up for the extra room).
boot0cfg changes:
+ modify the code to recognise the new boot0 structure (with the
relocated options block to make room for the Volume id).
+ add two options, '-i xxxx-xxxx' to set the volume ID, -e c
to modify the character printed in case of bad input
PR: 127764 70531
Submitted by: Christoph Mallon (portions)
MFC after: 4 weeks
of the boot0.S code, with a number of compile-time selectable options,
the most interesting one being the ability to select PXE booting.
The code is completely compatible with the previous one, and with
the boot0cfg program. Even the actual code is largely unmodified,
with only minor rearrangements or fixes to make room for the new
features.
The behaviour of the standard build differs from the previous
version in the following, minor things:
+ 'noupdate' is the default, which means the code does not
write back the selection to disk. You can enable the feature
at runtime with boot0cfg, or changing the flags in the Makefile.
+ a drive number of 0x00 (floppy, or USB in floppy emulation) is
now accepted as valid. Previously, it was overridden with 0x80,
meaning that the partition table coming from the media was
used to access sectors on a possibly different media.
You can revert to the previous mode building with -DCHECK_DRIVE,
and you can always use the 'setdrv' option in boot0cfg
+ certain FAT or NTFS partitions are listed as WIN instead of DOS.
+ the 'bel' character on a bad selection is replaced by a '#' to
make it clear that the system is not hang even if the machine
does not have a speaker. This can be reverted back at compile
time, or at runtime with an upcoming boot0cfg option.
Additional features are available as compile time options,
and may be become the default if deemed useful. In particular:
+ INT18/PXE boot (make -DPXE)
This option enables booting through INT 18h (which on certain
BIOSes can be hooked to PXE) by pressing F6. There is unfortunately
no room to print the additional menu option.
Also, to make room for the code, the 'Default: ' string is
changed to 'Boot: '
+ print current drive number (make -DTEST)
Prints a line indicating the current drive number.
This is useful to figure out what is going on for machines/bioses
which remap drives in sometimes surprising ways.
+ disable numeric keys in console mode (make -DONLY_F_KEYS)
Not really a significant option, but it is needed to make
room for the -DTEST mode.
+ disable floppy support (make -DCHECK_DRIVE)
Revert to the old behaviour of only accepting 0x80 and above
as valid drive numbers.
MFC after: 6 weeks
instead of "puts" which prints whatever is at %si, followed by a CRLF.
It was not noticed during tests because at that point %si points
to a partition entry whose first byte is 0x80, which is both a
terminator for the string and a non printable character.
Submitted by: Christoph Mallon
boot code. The bug was introduced in rev.1.13, and went unnoticed
because FreeBSD's boot1 does not use it, but other systems might.
(I have been struggling for almost a full day trying to figure out
why a syslinux'ed partition would not boot when started with the
FreeBSD /boot/boot0, only to realize that the bug was ours!)
The space for the two extra bytes (push %si and pop %si) is reclaimed
by removing an extra CRLF that is printed before booting.
The bug is not a major one but if there is time it might be a good
thing to merge it into the upcoming releases.
above) exhibits some misbehaviours on machines with AMD64 CPUs,
which at least in some cases I have tracked down to a heap overflow.
It is unclear whether it depends on the CPU or on the pxe bios
itself which may use more memory on AMD machines.
Noticeably a pxeboot compiled from 6.x sources works fine on all
machines I have tried so far, while a pxeboot compiled from 7.x
sources does not.
This patch is a first step in reducing the amount of memory used
while processing the configuration files read by the loader at boot
(some of them are quite large, 1700+ lines), and it does so by:
+ moving a buffer to static memory instead of allocating in the heap;
+ skipping empty lines;
+ reducing the amount of memory used for line descriptors;
Unfortunately there are several changes between 6.x and above,
affecting the compiler, the loader code itself, and libstand,
and it is not so straightforward to
These changes fix the behaviour on one motherboard with a
single-core AMD cpu, but are still not enough e.g on an Asus
M2N-VM (with a dual-core CPU).
I need to investigate the problem a bit more before figuring
out what should be committed to RELENG_7
PR: kern/118222
- Only non-sliced bsdlabel style partitioning is currently supported (but provisions
are made towards GPT support, which should follow soon)
- Enable storage support in loader on ARM
Obtained from: Semihalf
to gptboot, i.e. installed in a freebsd-boot partition using /sbin/gpart or
/sbin/gpt.
Tweak the /boot/loader ZFS support so that it can find ZFS pools that are
contained in GPT partitions.
This bring huge amount of changes, I'll enumerate only user-visible changes:
- Delegated Administration
Allows regular users to perform ZFS operations, like file system
creation, snapshot creation, etc.
- L2ARC
Level 2 cache for ZFS - allows to use additional disks for cache.
Huge performance improvements mostly for random read of mostly
static content.
- slog
Allow to use additional disks for ZFS Intent Log to speed up
operations like fsync(2).
- vfs.zfs.super_owner
Allows regular users to perform privileged operations on files stored
on ZFS file systems owned by him. Very careful with this one.
- chflags(2)
Not all the flags are supported. This still needs work.
- ZFSBoot
Support to boot off of ZFS pool. Not finished, AFAIK.
Submitted by: dfr
- Snapshot properties
- New failure modes
Before if write requested failed, system paniced. Now one
can select from one of three failure modes:
- panic - panic on write error
- wait - wait for disk to reappear
- continue - serve read requests if possible, block write requests
- Refquota, refreservation properties
Just quota and reservation properties, but don't count space consumed
by children file systems, clones and snapshots.
- Sparse volumes
ZVOLs that don't reserve space in the pool.
- External attributes
Compatible with extattr(2).
- NFSv4-ACLs
Not sure about the status, might not be complete yet.
Submitted by: trasz
- Creation-time properties
- Regression tests for zpool(8) command.
Obtained from: OpenSolaris
controller. The controller is also known as L1E(AR8121) and
L2E(AR8113/AR8114). Unlike its predecessor Attansic L1,
AR8121/AR8113/AR8114 uses completely different Rx logic such that
it requires separate driver. Datasheet for AR81xx is not available
to open source driver writers but it shares large part of Tx and
PHY logic of L1. I still don't understand some part of register
meaning and some MAC statistics counters but the driver seems to
have no critical issues for performance and stability.
The AR81xx requires copy operation to pass received frames to upper
stack such that ale(4) consumes a lot of CPU cycles than that of
other controller. A couple of silicon bugs also adds more CPU
cycles to address the known hardware bug. However, if you have fast
CPU you can still saturate the link.
Currently ale(4) supports the following hardware features.
- MSI.
- TCP Segmentation offload.
- Hardware VLAN tag insertion/stripping with checksum offload.
- Tx TCP/UDP checksum offload and Rx IP/TCP/UDP checksum offload.
- Tx/Rx interrupt moderation.
- Hardware statistics counters.
- Jumbo frame.
- WOL.
AR81xx PCIe ethernet controllers are mainly found on ASUS EeePC or
P5Q series of ASUS motherboards. Special thanks to Jeremy Chadwick
who sent the hardware to me. Without his donation writing a driver
for AR81xx would never have been possible. Big thanks to all people
who reported feedback or tested patches.
HW donated by: koitsu
Tested by: bsam, Joao Barros <joao.barros <> gmail DOT com >
Jan Henrik Sylvester <me <> janh DOT de >
Ivan Brawley < ivan <> brawley DOT id DOT au >,
CURRENT ML
on G4 machines. On the assumption that most people using FreeBSD on Apple
hardware are not using serial consoles, set boot1's output to screen. This
should be revisited. While here, reduce verbosity of boot1.
