Jeremy Allison c98ebb3031 Fix delete on close semantics to match W2K. I (think:-) I understand it now :-). ...

Thanks to Nir Livni <nirl@cyber-ark.com> for giving me the test case to
track it down.
Jeremy.

-

..

.cvsignore

add *.po32 to ignore list

-

blocking.c

Fixed nasty bug where file writes with start offsets in the range

-

build_options.c

BIG patch...

-

change_trust_pw.c

change_trust_account_password() must always use the PDC for rpc

-

chgpasswd.c

Removed level 0 DEBUG leftover from testing.

-

close.c

Fix delete on close semantics to match W2K. I (think:-) I understand it now :-).

-

conn.c

Non-error connection numbers are always positive

-

connection.c

BIG patch...

-

dfree.c

Removed version number from file header.

-

dir.c

Some cleanups:

-

dosmode.c

an initial fix for handling sparse files in smbd

-

error.c

Fix to correctly return NT_STATUS_DELETE_PENDING.

-

fileio.c

The element in fsp->print_job should be a RAP jobid, not a uint32 RPC

-

filename.c

make sure we don't run over the end of 'name' in unix_convert()

-

files.c

Fix for systems that allow more than 65536 open files per process.

-

ipc.c

Stop using hardcoded transact commands

-

lanman.c

Remove an unused paramter for our old LM-only password change code, and fix a

-

mangle_hash2.c

add a 'mangle prefix' option to allow people to tune the number of

-

mangle_hash.c

small merges from SAMBA_3_0; mostly typos, renames, etc...

-

mangle_map.c

Kill off const warnings - add a pile of const to various places.

-

mangle.c

BIG patch...

-

message.c

Address the string_sub problem by changing len = 0 to mean "no expand".

-

negprot.c

BIG patch...

-

noquotas.c

Removed version number from file header.

-

notify_hash.c

Removed version number from file header.

-

notify_kernel.c

Ensure we do not pass uninitialised data to the kernel. (Picked up by

-

notify.c

Added "kernel change notify" boolean to allow easier valgrind testing.

-

nttrans.c

BIG patch...

-

open.c

Ensure we only ever set fsp->conn in one place.

-

oplock_irix.c

Merge of receive_local_message fix from 2.2.5.

-

oplock_linux.c

Ensure we do not pass uninitialised data to the kernel. (Picked up by

-

oplock.c

Fix for interesting resource constraint condition. When all opens are

-

password.c

Some cleanups:

-

pipes.c

make sure we don't walk past the end of the current SMB buffer when

-

posix_acls.c

Patch from Edmund Lam <epl@unimelb.edu.au> to fix braindead Tru64 behaviour:

-

process.c

Add 3 second timeout when terminating server and sending print notify

-

quotas.c

performance patch from HP-UX folks (cant remember who)

-

reply.c

Make the 'service' in make_connection() use an fstrcpy(), and an fstring,

-

sec_ctx.c

Move to common user token debugging, and ensure we always print both the

-

server.c

Check for too many processes *before* the fork.

-

service.c

Oops, this is the change to use an fstring for the incoming service buffer -

-

session.c

Make 'remote_machine' private to lib/substitute.c, and fix all the user to use

-

sesssetup.c

Some cleanups:

-

srvstr.c

make sure we don't walk past the end of the current SMB buffer when

-

statcache.c

Some cleanups:

-

trans2.c

Fix delete on close semantics to match W2K. I (think:-) I understand it now :-).

-

uid.c

Removed global_myworkgroup, global_myname, global_myscope. Added liberal

-

utmp.c

BIG patch...

-

vfs-wrap.c

Actually use sendfile if selected.

-

vfs.c

untest fix for wide links parameter bug.

-