| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
| |
--with-lockdir and --with-privatedir.
JEremy.
|
|
|
|
|
|
|
|
|
| |
Upped MAX_OPEN_DIRECTORIES to 256.
Ensured new directory handles are only allocated between 255 - MAX.
Stopped re-open of a directory about to be closed.
Log a message if a directory is to be closed and the bitmap
is not set.
Jeremy.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
I have added a new define for configure - HAVE_EXPLICIT_LARGEFILE_SUPPORT.
This is tested for at the end of configure by the following check :
"As a gating factor for large file support, in order to
use <4GB files we must have the following minimal support
available.
long long, a 64 bit off_t or off64_t, a 64 bit ino_t or
ino64_t. If we don't have all of these then disable large
file support."
I have to do this as HPUX will detect the open64 etc.
functions in libc but *THEY DON'T WORK* before HPUX11.x.
The definitions in include/includes.h that set up large
file defines, and the calls in lib/system.c that use the explicit
large file functions are now depended on HAVE_EXPLICIT_LARGEFILE_SUPPORT
also being defined.
include/rpc_lsa.h: Upped number of groups to 96 as a temporary
measure.
Jeremy.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
was in seconds. This meant that PDC -> BDC failover took so long it essentiually
didn't happen.
include/client.h: Added explicit comment that cli->timeout should be milliseconds.
include/local.h: Changed LONG_CONNECT_TIMEOUT & SHORT_CONNECT_TIMEOUT to be milliseconds.
lib/smbrun.c: Tidied up fork failure catching.
lib/util_sock.c: Ensured timeout treated as milliseconds.
libsmb/clientgen.c: Added comment.
rpc_server/srv_pipe_hnd.c: Luke's changes.
smbd/chgpasswd.c: Paranoia code for EINTR around sys_waitpid.
Jeremy.
|
|
|
|
| |
Jeremy.
|
|
|
|
|
| |
the non-trans2 call - if you set the return parameters right.
Jeremy.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
problem.
printing/print_svid.c: Fix for DG/UX lpstat.
smbd/dfree.c: Added 'max_four_gig' parameter to normalize to 4GB when asked
(used to be the default).
smbd/open.c smbd/reply.c smbd/trans2.c:
Changed to allow a <4GB volume size to be returned on a trans2
qfsinfo level 1 call. NT clients refuse to do the correct call to Samba servers
(Herb recons it's the unicode bit) - this allows even a level 1 to return a
volume size up to 9444732961341243916800 bytes (should be enough for now,
until we get the unicode support added :-).
Jeremy.
|
|
|
|
| |
Jeremy.
|
|
|
|
| |
Jeremy.
|
|
|
|
| |
Added a function cli_set_port to clientgen.c
|
|
|
|
|
|
|
|
|
| |
--with-lockdir to
allow HP to specify where they want the /usr/local/samba/var/locks directory to go.
local.h: Added DEFAULT_PASSWD_CHAT to allow it to be changed.
param/loadparm.c: Use DEFAULT_PASSWD_CHAT and also re-arranged HPUX printing.
Jeremy.
|
|
|
|
|
|
| |
Outinen <Panu.Outinen@vertex.fi>
Jeremy.
|
|
|
|
|
|
| |
directory handle. WinCenter (multi-user NT system) is doing
this. This code still needs testing....
Jeremy.
|
|
|
|
| |
Jeremy.
|
|
|
|
|
|
| |
if you send this query to a FAT volume. Phew - I'm glad I got this one in
under the wire for 2.0.0.....
Jeremy.
|
|
|
|
|
|
|
|
| |
than it should :-).
SMBsearch now gets directory handles that are 8 bit in the range
1-255 and will close the oldest non-expect-close handle if the table gets full.
findfirst gets 16 bit directory handles starting at 256 and above.
Jeremy.
|
|
|
|
|
|
|
| |
time calculations, not by incrementing a counter.
The select timeout is set to the old (lower) value of 10 seconds if there
are blocking locks pending, and to the newer 60 second value if not.
Jeremy.
|
|
|
|
|
|
|
|
|
|
|
| |
beta5. The problem was that I had increased the select timeout so much
that the client was sending echo requests as a sort of keepalive.
As smbd only did timeout processing on select timeout then not
much timeout processing was being done :-).
I've fixed this by moving the timeout code to a function and
calling it every time we get an SMBecho request (we must have
been idle at that point).
Jeremy.
|
|
|
|
|
|
|
|
| |
array for directory handles. This tidied up the code a *lot*.
Plus it means I've fixed the bug in PR#10058, which was due to an
NT app that wanted to simultaneously open >256 directory handles.
Jeremy.
|
|
|
|
|
|
| |
Fixed <rpc/rpc.h> include bug for SCO.
Test for <arpa/inet.h> include for HPUX7.x.
Jeremy.
|
|
|
|
|
|
|
|
|
| |
--with-mmap code.
lib/time.c: Changed %T to %H:%M:%S for systems that don't have %T.
locking/locking_slow.c: Re-added FTRUNCATE_NEEDS_ROOT code - now to test for it...
smbd/quotas.c: Fixed IRIX quota compile.
Jeremy.
|
|
|
|
|
|
|
| |
source/include/local.h: Changed select look timeouts to 60 seconds (120 to
scan smb.conf file).
source/include/version.h: Updated to beta5.
Jeremy.
|
|
|
|
|
|
| |
All others - added checks for the compiler understanding volatile and
the system having sig_atomic_t. Changed sighup code to use this.
Jeremy.
|
|
|
|
| |
Jeremy.
|
|
|
|
|
|
|
| |
from Tim Rice.
utils/smbpasswd.c web/swat.c: Added set_auth_parameters() call where needed for OSF1_ENC_SEC.
Jeremy.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
smbd/chgpasswd.c.
smbd/chgpasswd.c
smbd/open.c: Ensure that we replace SIGCLD handler with CatchChildLeaveStatus()
*before* we expect to reap a child status - and reset it to CatchChild()
after we've done the sys_waitpid(). The original CatchChild() code
would eat the status code from the zombie in the signal handler, meaning
that the sys_waitpid() calls in check_access_allowed_for_current_user()
and chgpasswd would never return a status.
smbd/open.c:
smbd/reply.c:
smbd/close.c: Check error return on close() calls - ensure a valid error return
is sent back to the client. This catches the "file truncated"
bug when a user has run out of quota (or space) and the ENOSPC
error is returned on the *close()* call, not the write() (eg. on
an nfs mounted drive).
smbd/fileio.c: Added check for ESPIPE on sys_lseek() in file_seek() to allow
people to use Windows clients to communicate with UNIX fifo's
(like they could in 1.9.18).
Jeremy.
|
|
|
|
| |
Jeremy.
|
|
|
|
|
|
|
|
|
|
| |
Changed privilage to privilege.
Also added set_real_uid() call to lib/util_sec.c.
Removed last set[re]uid calls from quotas.c - all such calls now live only
in lib/util_sec.c.
Jeremy.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
All setuid code now resides in the one module lib/util_sec.c.
The interfaces this module exports are :
void gain_root_privilage(void); - Set real/eff/saved uid's to 0.
void gain_root_group_privilage(void); - Set real/eff/saved gid's to 0.
int set_effective_uid(uid_t uid); - Set eff uid *only* to given value.
int set_effective_gid(gid_t gid); - Set eff gid *only* to given value.
BOOL become_user_permanently(uid_t uid, gid_t gid); - Set real/eff/saved uid's and gid's
to uid and gid permanently - with no way back to root.
Most of the quota code now uses these calls (except for a few
special cases).
smbd/chgpasswd.c: Ensured the dochild exits in the fork()'d child.
libsmb/nmblib.c: Fix from Jasper for memory leak.
Jeremy.
|
|
|
|
|
|
| |
Fixed 'nmbd running wild' bug by preventing recursion in retransmit_or_expire_response_records().
tests/crypttest.c: Updated for bigcrypt.
Jeremy.
|
| |
|
|
|
|
|
| |
web/swat.c: Re-arranged static function to remove compile warning.
Jeremy.
|
|
|
|
|
| |
swat.c fixed create new share and printer
proto.h new function in loadparm.c
|
|
|
|
|
|
|
|
|
|
|
| |
back,
based on 1.9.18 tree.
tests/crypttest.c: Added test for truncated salt in crypt.
passdb/ldap.c: Malloc checks.
passdb/pass_check.c: Added test for truncated salt in crypt. Added OSF1 enhanced
security fix.
Jeremy.
|
|
|
|
|
|
|
|
| |
lib/system.c script/mkproto.awk include/includes.h: Added sys_readdir that
wraps readdir64.
lib/util.c: Changed gethostname() to use sizeof(hostname) as the second arg
for buffer size.
Jeremy.
|
|
|
|
|
| |
on Linux force fcntl/mmap based shared memory and on other systems
force IPC/shmem based shared memory.
|
|
|
|
|
|
| |
source/include/DesktopDB.h source/lib/streams.c: MAC support won't make it into 2.0,
although 2.0.1 will have it.
Jeremy.
|
|
|
|
|
|
|
| |
on share page
loadparm.c added FLAG_SHARE and FLAG_PRINT to various parameters to
tell swat which page these belong to
swat.c only print parameters marked on printers and shares pages
|
|
|
|
| |
Jeremy.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
changes are bzero -> memset as the compiler on HPUX is very picky about
bzero (probably a strange define).
The significant changes are in configure.in, to fix the HPUX shadow.h
problem and also move the large file support to the correct place.
passdb/pass_check.c: Added includes for KRB5, DCE and AFS if requested in autoconf.
tests/trapdoor.c: Fix to use setresuid and setresgid if defined.
The rest of the changes are bzero -> memset.
Jeremy.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
enumerate printers
param/loadparm.c added parameter to lp_dump to limit number of items
dumped. Pulled real dump code into new function
lp_dump_one so it can be called by swat
utils/testparm.c extra parameter added for lp_dump call
web/statuspage.c got rid of extra > being printed when smbd not
running
web/swat.c allow auto loaded printers to be enumerated in
printer page
include/proto.h changed number of parameters in lp_dump and new
function lp_dump_one defined.
|
|
|
|
|
|
|
|
|
|
| |
allow/deny that could
silently deny access. Log them if they occur.
smbd/close.c smbd/oplock.c smbd/reply.c: Removed a global oplock variable (hurrah!).
Removal of the oplocks flags and any kernel oplocks is now done in
either close() or in reply_locking() by calling release_file_oplock().
Jeremy.
|
|
|
|
|
|
|
|
| |
rpc_parse/parse_samr.c: Removed unneeded '&'s.
smbd/blocking.c:
smbd/reply.c: Fix for NT4.x bug where it sends large file lock requests
*even if Samba did not nogitiate large file support* !
Jeremy.
|
|\ |
|
| |
| |
| |
| | |
Jeremy.
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
| |
| |
| |
| |
| | |
LsaLookupSids etc from within SamrQueryAliasMembers, for example.
fnum is now a parameter to client functions. thanks to mike black
for starting the ball rolling.
|