| Age | Commit message (Collapse) | Author |
|---|
|
nasty routing table issues. Mainly the pointer to the previous node.
OK henning@
|
|
|
|
|
|
okay millert@
|
|
Implement file locking in lockd. All the stuff is done in userland, using
fhopen() and flock(). This means that if you kill lockd, all locks will
be relased.
Shared locks are not handled efficiently, they're serialised in lockd when
they could be granted.
tested against debian etch (linux 2.6.18)
not fully functional yet in mixed NFSv2/v3 environments
further development in-tree as discussed w/ deraadt
|
|
from NetBSD
minor bump, discussed w/ deraadt
|
|
|
|
from NetBSD
|
|
|
|
ok deraadt@, marco@
|
|
call to strtod() with bounding check.
Discussed with pyr@ and otto@
ok otto@ deraadt@
|
|
ok marco djm
|
|
- continuation line at file start would segfault: make it change ginfo
as it should instead.
- empty continuation lines should be valid.
okay millert@
|
|
|
|
original scientific paper inspiring that technique.
Much help with English and nroff by jmc@, thanks.
|
|
Noted, ok otto@
|
|
Add "c-handle-special-brace" which does a blink and indent, rather
than just a selfinsert and indent
|
|
system for the bridge ID for stp. That worksfine unless you have
two bridges in the system that talkto the same neighboring systems
(switches), because the two bridges on the openbsd system would have the
same ID.
fix by only looking at interfaces part of the bridge and using the lowest
mac address of these. works fine because stpcan only be enabled on
IFT_ETHER interfaces so there is always at least one and we re-evaluate
every time an interface is added or deleted from the bridge.
diff was rotting in my tree for at least a year, I have no idea what
triggered it really.
ok reyk mk djm
|
|
function. While here, tighten whitespace.
|
|
|
|
If you use this, change your habits (and .mg) to use the following
commands instead:
mail-mode => auto-fill-mode
mail-set-margin => set-fill-column
ok henning@. Will also "fix" debian bug #460189
|
|
'Prints root directory as "//" instead of "/" for root files.'
Issue was with dirname, which strips the trailing slash, except
when given "/". Wrap it in a cover function to fix. Also helps
with portability to data-munging dirname glibc.
|
|
Two issues: *scratch* did not inherit any modes set from
~/.mg, as it was created too early. Switch to using *init* buffer
until after autoexec, then kill it and inherit correct *scratch*.
Second, -f should force the mode. Not toggle it.
ok phessler@
|
|
parts in #ifdef _KERNEL; makes softraidvar.h usable from userland
ok hshoexer@
|
|
mask key from userland. Thus, modify sr_crypto_get_kdf() do correclty handle
copying kdf hint and/or mask key.
While there, remove superflous sr_read_meta().
ok marco djm
|
|
other drivers.
|
|
instead of getting truncated.
|
|
|
|
|
|
ok blambert@
|
|
it again in that function.
|
|
On socppc it has the subclass PCI_SUBCLASS_MASS_STORAGE_MISC
so without the quirk socppc will not find its disk.
-moj
|
|
Problem found by Martynas Venckus, fix by me.
|
|
|
|
|
|
applying additional restrictions to non-pubkey authentications in
the case where pubkey fails but another method subsequently succeeds.
bz #1472, found by Colin Watson, ok markus@ djm@
|
|
|
|
|
|
ok @marco
|
|
|
|
not randomly skip probing for devices after a missing device.
ok kettenis
|
|
time computation.
|
|
From davidch @ FreeBSD
|
|
sure to #include "foo.h" in all code that uses the softc,
otherwise some object files will have a different idea of how
big the softc struct is.
This will result in things like part of the softc, say the interrupt
handler being overridden by other data. Putting printfs in the
main object file will look fine as it has the right idea as to
what the size is. It is only after several hours debugging
that you start to figure out what the hell is going on.
jsing started trying to figure out the problem and got hints
from kettenis that it likely wasn't an interrupt problem,
after quite a bit of head scratching with joel I eventually spotted
the problem.
ok reyk@ jsing@
|
|
include the program name in logmsg() plus exit_daemon() does not return so
make it a void function. OK mpf@
|
|
|
|
That will be sizeof(long) no matter what.
|
|
This restores the old behaviour where we were checking for != LINK_STATE_DOWN.
Fixes the ethernet to wireless failover feature and also allows trunk(4)
to be used within qemu.
Tested by sturm@.
OK reyk@. "grumble, we should fix more drivers' link state handling" brad@
|
|
as host byte order in userland. ifconfig didn't get this and always printed
the pfsync syncpeer on little endian machines because the check to prevent
printing the default address assumed the wrong byte order.
ok claudio@ rainer@
|
|
it once at start instead.
|
