being defined, define and use a new MD macro, cpu_spinwait(). It only
expands to something on i386 and amd64, so the compiled code should be
identical.
Name of the macro found by: jhb
Reviewed by: jhb
make it fully self-contained.
o ip_reass() now returns a new mbuf with the reassembled packet and ip->ip_len
including the IP header.
o Computation of the delayed checksum is moved into divert_packet().
Reviewed by: silby
- according to RFC2661 an offset size of 0 is allowed.
- when skipping offset padding do not forget to also skip
the 2 octets of the offset size field.
Reviewed by: archie
Approved by: pjd (mentor)
link[n].latency calculated from user supplied value.
This prevents repeated NGM_PPP_SET_CONFIG/NGM_PPP_GET_CONFIG
from failing because of link[n].conf.latency being out of range.
Reviewed by: archie
Approved by: pjd (mentor)
and setting MSR. This was most evident with the idle proc running
with interrupts disabled and causing a lockup. Switch over to the
i386 style which does things in the right order.
debug assisted by: gallatin, and the invaluable KTR option.
pipelock(), not via a mixture of mutexes and pipelock(). Additionally,
add a few KASSERTS, and change some statements that should have been
KASSERTS into KASSERTS.
As a result of these cleanups, some segments of code have become
significantly shorter and/or easier to read.
unconditionally, stop after the first one (system board) if no EISA hardware
is detected. This fixes a boot hang (i.e. Thinkpad) when ACPI is disabled.
Also, split the probe code into a separate function and do some style cleanup.
Note that the Adaptec 2842 VLB controller probe is broken by this change
and will fail to probe. It should be fixed separately.
in case of a CHECK CONDITION.
- Make this driver return SCSI status information.
- While here, factor out the clearing of the CAM status from every
element of the switch statement to only once before the switch.
This fixes burning CDs with recent cdrecord 2.01 alpha versions and
burners attached to asr(4) controllers but there could have been
other applications and da(4) etc. also affected.
Reviewed by: gibbs, scottl
MFC after: 2 weeks
UFS2 was here. It so happened that UFS2 did not need a seperate
partition type. Keep the definition as a comment for documentation
purposes. If there is a benefit for UFS2 file systems to have a
seperate partition type under GPT, then this definition should be
restored as that was the intention of the definition.
Currently one cannot load the mem.ko module without panicing if mem is
compiled into the kernel and one cannot build a kernel w/o "device mem"
right now either. Thus it is too dangerous to install mem.ko right now
because if one puts 'mem_load="YES"' in /etc/loader.conf they cannot
boot an "old" kernel (at the time that a kernel doesn't have to be built
with "device mem).
pic_eoi_source() into one call. This halves the number of spinlock operations
and indirect function calls in the normal case of handling a normal (ithread)
interrupt. Optimize the atpic and ioapic drivers to use inlines where
appropriate in supporting the intr_execute_handlers() change.
This knocks 900ns, or roughly 1350 cycles, off of the time spent servicing an
interrupt in the common case on my 1.5GHz P4 uniprocessor system. SMP systems
likely won't see as much of a gain due to the ioapic being more efficient than
the atpic. I'll investigate porting this to amd64 soon.
Reviewed by: jhb
skip blocks that are too big by a factor of two or greater. This
avoids some cases of extremely inefficient memory use that can occur
when large (e.g. 64k) blocks on the free list get used when allocating
a 4k chunk of 64-byte fragments. Because fragments have their own
free list, the 60k difference got lost forever every time.
system BIOS to disable legacy device emulation as per the "EHCI
Extended Capability: Pre-OS to OS Handoff Synchronisation" section
of the EHCI spec. BIOSes that implement legacy emulation using SMIs
are supposed to disable the emulation when this procedure is performed.
set gp->softc to NULL and return ENXIO when it is NULL, so GEOM
will not panic or hang, but unload one device on every 'unload'.
This make 'unload' command usable, but it have to be executed
<number of devices> + 1 times.
- Made use of 'pp' variable.
so that they know whether the allocation is supposed to be able to sleep
or not.
* Allow uma_zone constructors and initialation functions to return either
success or error. Almost all of the ones in the tree currently return
success unconditionally, but mbuf is a notable exception: the packet
zone constructor wants to be able to fail if it cannot suballocate an
mbuf cluster, and the mbuf allocators want to be able to fail in general
in a MAC kernel if the MAC mbuf initializer fails. This fixes the
panics people are seeing when they run out of memory for mbuf clusters.
* Allow debug.nosleepwithlocks on WITNESS to be disabled, without changing
the default.
Both bmilekic and jeff have reviewed the changes made to make failable
zone allocations work.
now, but it's possible for ndis_reset_nic() to sleep (sometimes the
MiniportReset() method returns NDIS_STATUS_PENDING and we have
to wait for completion). To get around this, execute the ndis_reset_nic()
routine in the NDIS_TASKQUEUE thread.
- Give ndiscvt(8) the ability to process a .SYS file directly into
a .o file so that we don't have to emit big messy char arrays into
the ndis_driver_data.h file. This behavior is currently optional, but
may become the default some day.
- Give ndiscvt(8) the ability to turn arbitrary files into .ko files
so that they can be pre-loaded or kldloaded. (Both this and the
previous change involve using objcopy(1)).
- Give NdisOpenFile() the ability to 'read' files out of kernel memory
that have been kldloaded or pre-loaded, and disallow the use of
the normal vn_open() file opening method during bootstrap (when no
filesystems have been mounted yet). Some people have reported that
kldloading if_ndis.ko works fine when the system is running multiuser
but causes a panic when the modile is pre-loaded by /boot/loader. This
happens with drivers that need to use NdisOpenFile() to access
external files (i.e. firmware images). NdisOpenFile() won't work
during kernel bootstrapping because no filesystems have been mounted.
To get around this, you can now do the following:
o Say you have a firmware file called firmware.img
o Do: ndiscvt -f firmware.img -- this creates firmware.img.ko
o Put the firmware.img.ko in /boot/kernel
o add firmware.img_load="YES" in /boot/loader.conf
o add if_ndis_load="YES" and ndis_load="YES" as well
Now the loader will suck the additional file into memory as a .ko. The
phony .ko has two symbols in it: filename_start and filename_end, which
are generated by objcopy(1). ndis_open_file() will traverse each module
in the module list looking for these symbols and, if it finds them, it'll
use them to generate the file mapping address and length values that
the caller of NdisOpenFile() wants.
As a bonus, this will even work if the file has been statically linked
into the kernel itself, since the "kernel" module is searched too.
(ndiscvt(8) will generate both filename.o and filename.ko for you).
- Modify the mechanism used to provide make-pretend FASTCALL support.
Rather than using inline assembly to yank the first two arguments
out of %ecx and %edx, we now use the __regparm__(3) attribute (and
the __stdcall__ attribute) and use some macro magic to re-order
the arguments and provide dummy arguments as needed so that the
arguments passed in registers end up in the right place. Change
taken from DragonflyBSD version of the NDISulator.
to be particularly correct or optimal, but it seems to be enough
to allow the attachment of USB2 hubs and USB2 devices connected via
USB2 hubs. None of the split transaction support is implemented in
our USB stack, so USB1 peripherals will definitely not work when
connected via USB2 hubs.
their own directory and module, leaving the MD parts in the MD
area (the MD parts _are_ part of the modules). /dev/mem and /dev/io
are now loadable modules, thus taking us one step further towards
a kernel created entirely out of modules. Of course, there is nothing
preventing the kernel from having these statically compiled.
The different between the new function and g_mirror_orphan() (which was
used previously) is that syncid is bumped immediately, instead of on
first write, because when consumer was spoiled, it means, that its
provider was opened for writing, so we can't trust that its data
will be valid when it will be connected again.
features. The gmirror(8) utility should be used for control of this class.
There is no manual page yet, but I'm working on it with keramida@.
Many useful tests provided by: simon (thank you!)
Some ideas from: scottl, simon, phk
and refuse initializing filesystems with a wrong version. This will
aid maintenance activites on the 5-stable branch.
s/vfs_mount/vfs_omount/
s/vfs_nmount/vfs_mount/
Name our filesystems mount function consistently.
Eliminate the namiedata argument to both vfs_mount and vfs_omount.
It was originally there to save stack space. A few places abused
it to get hold of some credentials to pass around. Effectively
it is unused.
Reorganize the root filesystem selection code.
those architectures without pmap locking.
- Eliminate the acquisition and release of Giant from vm_map_protect().
(Translation: mprotect(2) runs to completion without touching Giant on
alpha, amd64, i386 and ia64.)
brings ia64 to parity with alpha, amd64, and i386 in this area.)
- Prevent a race in pmap_find_pte(): If pmap_find_pte() sleeps in
uma_zalloc(), another thread could allocate a pte at the same address.
Instead, sleep at a higher level and retry the lookup before retrying
the allocation.
Reviewed and tested by: marcel@
This is really ugly way to do this, but there is no other way for now.
It allows to mount root file system from providers which belong to
those classes.
Approved by: phk
maps. We always acquire the sx lock exclusively here, but we can't
use a mutex because we want to be able to sleep while holding the
lock. This is completely equivalent to what we were doing with the
lockmgr(9) locks before.
Approved by: alc
provider.
- Bump version number.
This allows for a quite interesting trick. One can setup a stripe with
stripe size of 512 bytes and create transparent provider on top of it
with sector size equal to <ndisks> * 512. The result will be something
like RAID3 without parity disk (every access will touch all disks).
submitted version with style cleanups and changes to comments. I also
modified the ioctl interface. This version only has one ioctl (to get
the Synaptics-specific config parameters) since this is the only
information a user might want.
Submitted by: Arne Schwabe <arne -at- rfc2549.org>
- Enable recursion on the page queues lock. This allows calls to
vm_page_alloc(VM_ALLOC_NORMAL) and UMA's obj_alloc() with the page
queues lock held. Such calls are made to allocate page table pages
and pv entries.
- The previous change enables a partial reversion of vm/vm_page.c
revision 1.216, i.e., the call to vm_page_alloc() by vm_page_cowfault()
now specifies VM_ALLOC_NORMAL rather than VM_ALLOC_INTERRUPT.
- Add partial locking to pmap_copy(). (As a side-effect, pmap_copy()
should now be faster on i386 SMP because it no longer generates IPIs
for TLB shootdown on the other processors.)
- Complete the locking of pmap_enter() and pmap_enter_quick(). (As of now,
all changes to a user-level pmap on alpha, amd64, and i386 are performed
with appropriate locking.)
- zone_large_init() stays pretty much the same.
- zone_small_init() will try to stash the slab header in the slab page
being allocated if the amount of calculated wasted space is less
than UMA_MAX_WASTE (for both the UMA_ZONE_REFCNT case and regular
case). If the amount of wasted space is >= UMA_MAX_WASTE, then
UMA_ZONE_OFFPAGE will be set and the slab header will be allocated
separately for better use of space.
- uma_startup() calculates the maximum ipers required in offpage slabs
(so that the offpage slab header zone(s) can be sized accordingly).
The algorithm used to calculate this replaces the old calculation
(which only happened to work coincidentally). We now iterate over
possible object sizes, starting from the smallest one, until we
determine that wastedspace calculated in zone_small_init() might
end up being greater than UMA_MAX_WASTE, at which point we use the
found object size to compute the maximum possible ipers. The
reason this works is because:
- wastedspace versus objectsize is a see-saw function with
local minima all equal to zero and local maxima growing
directly proportioned to objectsize. This implies that
for objects up to or equal a certain objectsize, the see-saw
remains entirely below UMA_MAX_WASTE, so for those objectsizes
it is impossible to ever go OFFPAGE for slab headers.
- ipers (items-per-slab) versus objectsize is an inversely
proportional function which falls off very quickly (very large
for small objectsizes).
- To determine the maximum ipers we'll ever need from OFFPAGE
slab headers we first find the largest objectsize for which
we are guaranteed to not go offpage for and use it to compute
ipers (as though we were offpage). Since the only objectsizes
allowed to go offpage are bigger than the found objectsize,
and since ipers vs objectsize is inversely proportional (and
monotonically decreasing), then we are guaranteed that the
ipers computed is always >= what we will ever need in offpage
slab headers.
- Define UMA_FRITM_SZ and UMA_FRITMREF_SZ to be the actual (possibly
padded) size of each freelist index so that offset calculations are
fixed.
This might fix weird data corruption problems and certainly allows
ARM to now boot to at least single-user (via simulator).
Tested on i386 UP by me.
Tested on sparc64 SMP by fenner.
Tested on ARM simulator to single-user by cognet.
* Some systems have _FDE and child floppy devices, but no _FDI. This seems
to be compatible with the standard. Don't error out if there is no _FDI.
Instead, continue on to the next device. The normal fd probe will take
care of this device.
* Some systems have _FDE but no child devices in AML. For these, add a
second pass that compares the results of _FDE to the presence of devices.
If not present, add the missing device.
* Some BIOS authors didn't read the spec. They use tape drive values for
all fdc(4) devices. Since this isn't grossly incompatible with the
required boolean value, use them. They also define the _FDE items as a
package instead of buffer. Regenerate the buffer from the package if it
is present.
Tested by: tjr, marcel
Add local rootvp variables as needed.
Remove checks for miniroot's in the swappartition. We never did that
and most of the filesystems could never be used for that, but it had
still been copy&pasted all over the place.
vm/vm_object.c revision 1.88) and vm_object_sync() (originating in
vm/vm_map.c revision 1.36): When descending a chain of backing objects,
both use the wrong object's backing offset. Consequently, both may
operate on the wrong pages.
Quoting Matt, "This could be responsible for all of the sporatic madvise
oddness that has been reported over the years."
Reviewed by: Matt Dillon
Alice is too lazy to write a server application in PF-independent
manner. Therefore she knocks up the server using PF_INET6 only
and allows the IPv6 socket to accept mapped IPv4 as well. An evil
hacker known on IRC as cheshire_cat has an account in the same
system. He starts a process listening on the same port as used
by Alice's server, but in PF_INET. As a consequence, cheshire_cat
will distract all IPv4 traffic supposed to go to Alice's server.
Such sort of port theft was initially enabled by copying the code that
implemented the RFC 2553 semantics on IPv4/6 sockets (see inet6(4)) for
the implied case of the same owner for both connections. After this
change, the above scenario will be impossible. In the same setting,
the user who attempts to start his server last will get EADDRINUSE.
Of course, using IPv4 mapped to IPv6 leads to security complications
in the first place, but there is no reason to make it even more unsafe.
This change doesn't apply to KAME since it affects a FreeBSD-specific
part of the code. It doesn't modify the out-of-box behaviour of the
TCP/IP stack either as long as mapping IPv4 to IPv6 is off by default.
MFC after: 1 month
this file from userland. Since we export struct ifnet to userland, and
that struct ifnet now contains a struct task, userland needs to know
what struct task looks like.
We need to consider having a pointer to a struct task here instead and
forward declare struct task in the !_KERNEL case.
static symbols. This wasn't a problem with previous GCC releases, but
unit-at-a-time mode of GCC 3.4.2 prevents linker set components from
being emitted at all.
"__FreeBSD_version should only ever increment. It is a historial record
of events in the system. Decrementing it is akin to trying to go back
in time and change history."
Reminded by: kuriyama, scottl
with the FIN bit set for all segments, if a FIN has already been sent before.
The fix will allow the FIN bit to be set for only the last segment, in case
it has to be retransmitted.
Fix another bug that would have caused snd_nxt to be pulled by len if
there was an error from ip_output. snd_nxt should not be touched
during sack retransmissions.
synchronizing IPv6 protocol control blocks and lists. These changes
are modeled on the inpcb locking for IPv4, submitted by Jennifer Yang,
and committed by Jeffrey Hsu. With these locking changes, IPv6 use of
inpcbs is now substantially more MPSAFE, and permits IPv4 inpcb locking
assertions to be run in the presence of IPv6 compiled into the kernel.
device drivers to declare that the ifp->if_start() method implemented
by the driver requires Giant in order to operate correctly.
Add a 'struct task' to 'struct ifnet' that can be used to execute a
deferred ifp->if_start() in the event that if_start needs to be called
in a Giant-free environment. To do this, introduce if_start(), a
wrapper function for ifp->if_start(). If the interface can run MPSAFE,
it directly dispatches into the interface start routine. If it can't
run MPSAFE, we're running with debug.mpsafenet != 0, and Giant isn't
currently held, the task is queued to execute in a swi holding Giant
via if_start_deferred().
Modify if_handoff() to use if_start() instead of direct dispatch.
Modify 802.11 to use if_start() instead of direct dispatch.
This is intended to provide increased compatibility for non-MPSAFE
network device drivers in the presence of Giant-free operation via
asynchronous dispatch. However, this commit does not mark any network
interfaces as IFF_NEEDSGIANT.
using linker_load_module(). This works OK if NGM_MKPEER message came
from userland and we have process associated with thread. But when
NGM_MKPEER was queued because target node was busy, linker_load_module()
is called from netisr thread leading to panic.
To workaround that we do not load modules by framework, instead ng_socket
loads module (if this is required) before sending NGM_MKPEER.
However, the race condition between return from NgSendMsg() and actual
creation of node still exist and needs to be solved.
PR: kern/62789
Approved by: julian
clients simultaneously. When node is client its mode is configured
with a control message.
sysctl net.graph.nonstandard_pppoe is deprecated but kept for
backward compatibility for some time.
Approved by: julian
dereference curthread. It is called only from critical_{enter,exit}(),
which already dereferences curthread. This doesn't seem to affect SMP
performance in my benchmarks, but improves MySQL transaction throughput
by about 1% on UP on my Xeon.
Head nodding: jhb, bmilekic
(i.e. with the foreign address being not wildcard) when checking
for possible port theft since such connections cannot be stolen.
The port theft check is FreeBSD-specific and isn't in the KAME tree.
PR: bin/65928 (in the audit trail)
Reviewed by: -net, -hackers (silence)
Tested by: Nick Leuta <skynick at mail.sc.ru>
MFC after: 1 month
an adaptive fashion when adaptive mutexes are enabled. The theory
behind non-adaptive Giant is that Giant will be held for long periods
of time, and therefore spinning waiting on it is wasteful. However,
in MySQL benchmarks which are relatively Giant-free, running Giant
adaptive makes an observable difference on SMP (5% transaction rate
improvement). As such, make adaptive behavior on Giant an option so
it can be more widely benchmarked.
- Push down Giant into shmexit(). (Giant is acquired only if the vmspace
contains shm segments.)
- Eliminate the acquisition of Giant from proc_rwmem().
- Reduce the scope of Giant in exit1(), uncovering the destruction of the
address space.
switch in fork_exit() to before anything else is done (but keep
schedlock for the deadthread check). This means one less
nasty bug if ever in the future whatever might have been called
before the update played with schedlock or critical sections.
Discussed with: tjr
when inpcb is NULL, this is no longer invalid since jlemon added the
tcp_twstart function... this prevents close "failing" w/ EINVAL when it
really was successful...
Reviewed by: jeremy (NetBSD)
Now it is user-controlled through ifconfig(8).
The former ``automagic'' way of operation created more
trouble than good. First, VLAN_MTU consumers other than
vlan(4) had appeared, e.g., ng_vlan(4). Second, there was
no way to disable VLAN_MTU manually if it were causing
trouble, e.g., data corruption.
Dropping the ``automagic'' should be completely invisible
to the user since
a) all the drivers supporting VLAN_MTU
have it enabled by default, and in the first place
b) there is only one driver that can really toggle VLAN_MTU
in the hardware under its control (it's fxp(4), to which
I added VLAN_MTU controls to illustrate the principle.)
the system" resource limit code: When checking if the caller has superuser
privileges, we should be checking the *real* user, not the *effective*
user. (In general, resource limiting is done based on the real user, in
order to avoid resource-exhaustion-by-setuid-program attacks.)
Now that a SUSER_RUID flag to suser_cred exists, use it here to return
this code to its correct behaviour.
Pointed out by: rwatson
somewhat clearer, but more importantly allows for a consistent naming
scheme for suser_cred flags.
The old name is still defined, but will be removed in a few days (unless I
hear any complaints...)
Discussed with: rwatson, scottl
Requested by: jhb
ACPI_DEBUG. This upset the ordering that acpi_probe_order() was meant to
provide, causing devices to attach before the sysresource object. This
debugging feature has been unnecessary for a while so just remove it.
Testing by: marcel
the license from /usr/src/COPYRIGHT. Since cvs annotate shows that
this was written by jasone, julian, jhb, peter, bmilekic and obrien.
cvs log shows that many others may have contributed to this file. As
such, go ahead and use the author of 'FreeBSD Project' for this file.
If this is a problem, please notify me.
# this eliminates the last file in the kernel with an indirect reference
# to /usr/src/COPYRIGHT in the kernel. A few more in userland remain.
vinumdrive geom with an exclusive bit. This should fix the problem
when underlying partitions overlap (i.e. the 'a' partition is at
the same offset as the 'c' partition).
Ideas borrowed from pjd@, quite a bit of testing by
Matthias Schuendehuette <msch@snafu.de>.
Properly wait for not busy and introduce a timeout for devices not
setting busy (as they should).
Leave a printf in there that states how long the wait was, as I'd like
to get an idea of the variations here. The time needed seems also to be
affected by whether a medium is present or not.
FOREACH_SAFE. Remove bad cast of retp and instead use an additional
arg to pass back the number of valid outputs. Use the package convenience
functions for parsing packages.
o Separate out local (ports) scripts that use rc.d, and the old style
startup/shutdown scripts and execute them separately. On startup the
rc.d style scripts are executed first and then the old-style scripts.
On shutdown, exactly the reverse happens.
o The rc.d ports scripts should now behave more like base system scripts.
Scripts ending in .sh will be sourced into the current shell, while the
rest will be executed in a subshell. Previously, all ports scripts,
regardless of the .sh suffix, were executed in a subshell.
o The parent script, /etc/rc.d/localpkg, passes its command line arguments
straight to the rc.d ports scripts. This means they should now honor
faststop and faststart commands as well. Old style scripts, should not see
any differences. They will still get either a start or stop command.
o The initial phrase shown during shutdown has been changed to use
"local packages" instead of "daemon processes" to be more inline with the
phrase used during local package startup. The phrases are also used only for
old-style ports script startup/shutdown, whereas previously they were being
used for both rc.d and old-style scripts. This should make startup/shutdown
output a bit less ugly.
Discussed with: portmgr
Has Reservations: eik
(in particular, bge(4) hasn't supported rxcsum since if_bge.c#1.5)
Clean up some aspects of capabilities usage, i.e. stop using
if_hwassist to see whether we are doing offload now because if_hwassist
is for TCP/IP layer and it is subordinate to if_capenable.
Thanks to: Aled Morris for donating a nice bge(4) NIC to me
Reviewed by: -net, -hackers (silence)
vmspace to the new vmspace in vmspace_exec() is mostly wasted effort. With
one exception, vm_swrss, the copied fields are immediately overwritten.
Instead, initialize these fields to zero in vmspace_alloc(), eliminating a
bcopy() from vmspace_exec() and a bzero() from vmspace_fork().
can't yet be referenced by other threads.
In microbenchmarks, this appears to reduce the cost of
pipe();close();close() on UP by 10%, and SMP by 7%. The vast majority
of the cost of allocating a pipe remains VM magic.
Suggested by: silby
calls further down the stack. If we find the cksum to be okay we pretend
that the hardware did all the work and hence keep the upper layers from
checking again.
Submitted by: Pyun YongHyeon
addend of 0. This isn't correct, and was quite easy to break by
referring to the address of an element within a structure.
However, fixing this exposed the fact that symbol lookups for
local variables were returning the base of the section they
were contained in. This case is detected by comparing the return
value from elf_lookup() to the relocbase+addend value: if it is
lesser, but greater than relocbase, then relocbase+addend is
taken to be the authoritative value.
bug reported by: gallatin
happens because the sio device was never opened and com->tp is
therefore NULL. ttygone can't swallow a NULL, so guard against that
possibility. Other places in this function make similar checks, so I
believe this is correct.
Improve child_detached a little and make it conform better to
style(9). Also, improve comment about what we'll be doing in the
future about driver_added. Soon it will be possible to kldload usb
drivers and have them attach w/o a need to disconnect/reconnect them.
Giant conditional on debug.mpsafenet in the socket soo_stat() routine,
unconditionally in vn_statfile() for VFS, and otherwise don't acquire
Giant. Accept an unlocked read in kqueue_stat(), and cryptof_stat() is
a no-op. Don't acquire Giant in fstat() system call.
Note: in fdescfs, fo_stat() is called while holding Giant due to the VFS
stack sitting on top, and therefore there will still be Giant recursion
in this case.
the data sheets leads me to believe these will just work. Those parts
with the various media readers on them may not have the required
FreeBSD drivers that will attach to the subdevices that will be seen
on some of these parts.
PCI 1515, 1530, 1620, 4520, 6411, 6420, 7410, 7510, 7610
Prompted by: Havard Eidnes
