| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
realloc can return NULL in one of two cases - (1) the realloc failed,
(2) realloc succeeded but the new size requested was zero, in which
case this is identical to a free() call.
The error paths dealing with these two cases should be different,
but mostly weren't. Secondly the standard idiom for dealing with
realloc when you know the new size is non-zero is the following :
tmp = realloc(p, size);
if (!tmp) {
SAFE_FREE(p);
return error;
} else {
p = tmp;
}
However, there were *many* *many* places in Samba where we were
using the old (broken) idiom of :
p = realloc(p, size)
if (!p) {
return error;
}
which will leak the memory pointed to by p on realloc fail.
This commit (hopefully) fixes all these cases by moving to
a standard idiom of :
p = SMB_REALLOC(p, size)
if (!p) {
return error;
}
Where if the realloc returns null due to the realloc failing
or size == 0 we *guarentee* that the storage pointed to by p
has been freed. This allows me to remove a lot of code that
was dealing with the standard (more verbose) method that required
a tmp pointer. This is almost always what you want. When a
realloc fails you never usually want the old memory, you
want to free it and get into your error processing asap.
For the 11 remaining cases where we really do need to keep the
old pointer I have invented the new macro SMB_REALLOC_KEEP_OLD_ON_ERROR,
which can be used as follows :
tmp = SMB_REALLOC_KEEP_OLD_ON_ERROR(p, size);
if (!tmp) {
SAFE_FREE(p);
return error;
} else {
p = tmp;
}
SMB_REALLOC_KEEP_OLD_ON_ERROR guarentees never to free the
pointer p, even on size == 0 or realloc fail. All this is
done by a hidden extra argument to Realloc(), BOOL free_old_on_error
which is set appropriately by the SMB_REALLOC and SMB_REALLOC_KEEP_OLD_ON_ERROR
macros (and their array counterparts).
It remains to be seen what this will do to our Coverity bug count :-).
Jeremy.
(This used to be commit 1d710d06a214f3f1740e80e0bffd6aab44aac2b0)
|
|
* only keep the registry,tdb file open when we have an open key handle
* tpot's setup.py fix
* removing files that no longer exist in trunk and copying some
that were missing in 3.0
(This used to be commit 6c6bf6ca5fd430a7a20bf20ed08050328660e570)
|
|
* \PIPE\unixinfo
* winbindd's {group,alias}membership new functions
* winbindd's lookupsids() functionality
* swat (trunk changes to be reverted as per discussion with Deryck)
(This used to be commit 939c3cb5d78e3a2236209b296aa8aba8bdce32d3)
|
|
(This used to be commit ef7e0d70c68976766c01e1212e2b8f48f6895f98)
|
|
HKLM\\SYSTEM\\CurrentControlSet\\Control\\Termininal Server\\DefaultUserConfiguration
Apparently this started showing up after the winreg-write support
was added in 3.0.20rc1 or so.
Also modifed init_registry_data() to always run and add the
required keys. Initial values however are only written if
they don't already exist.
This makes it easier to add new keys without having to rev the
tdb version number (which is really unnecessary in this case).
Portions of patch reviewed by Thomas Bork on the general samba ml.
(This used to be commit b12a05b23782cfcb93fb4811807ef388de97c95c)
|
|
the new talloc() features:
Note that the REGSUB_CTR and REGVAL_CTR objects *must* be talloc()'d
since the methods use the object pointer as the talloc context for
internal private data.
There is no longer a regXXX_ctr_intit() and regXXX_ctr_destroy()
pair of functions. Simply TALLOC_ZERO_P() and TALLOC_FREE() the
object.
Also had to convert the printer_info_2->NT_PRINTER_DATA field
to be talloc()'d as well. This is just a stop on the road to
cleaning up the printer memory management.
(This used to be commit ef721333ab9639cb5346067497e99fbd0d4425dd)
|
|
(This used to be commit 5d592691e4d4f1cd47bd062633a54d7892548ee7)
|
|
(This used to be commit e9427912a763b0e4bb47724f835b91c2252105e8)
|
|
for NT4 clients enumerating printer data on slow CPUs)
* fix pinter and secdesc record upgrade to normalize the key
(rev'd printer tdb version)
* fixed problem that was normalizing the printername name field
in general, this should fix the issues upgrading print servers
from 3.0.14a to 3.0.20
(This used to be commit d07179de2f2a6eb1d13d0e25ac10de1a21475559)
|
|
* use SAMBA_PRINTER_PORT_NAME in registry values for builtin printer
port
(This used to be commit 63bc03536b6d0622005448f0f7be2739e06a432a)
|
|
* more work on the store_values() functions for the Printers key
* add Control\Print\Monitors key to list for reg_db
(This used to be commit 89f17b41cee633838b8cbd0d1bf8119a4b8d707e)
|
|
* move to registry.tdb for port listing (at least via the winreg ops)
If no one opposes on the samba list, we'll move to a registry
lookup for enumerating ports rather than the 'enumports command'.
This means that there is a bit of a disconnect between EnumPorts() and
RegEnumKey('hklm\software\microsoft\windows nt\currentversion\ports').
(This used to be commit 6f654c5741e98abf00c010c5dd38038092c0f7a3)
|
|
(still not completely back to the read functionality
we previously had but the cleanup is progressing)
(This used to be commit 04431372a698433b4936392047228908a64ff382)
|
|
* make regdb_store_XXX() and regdb_fetch_XXX() functions non-static
* use case sensitive string lookups in reg_dynamic.c since the
keys have already been normalized
* move to new design for making printing related data available
via the winreg pipe (with the intent of allowing writes)
(This used to be commit 28c7293ee9e68b913faf8d74d63f73e09087169b)
|
|
printmig.exe assumes that the LUID of the SeBackupPrivlege
on the target server matches the LUID of the privilege
on the local client. Even though an LUID is never guaranteed
to be the same across reboots. How *awful*! My cat could
write better code! (more on my cat later....)
* Set the privelege LUID in the global PRIVS[] array
* Rename RegCreateKey() to RegCreateKeyEx() to better match MSDN
* Rename the unknown field in RegCreateKeyEx() to disposition
(guess according to MSDN)
* Add the capability to define REG_TDB_ONLY for using the reg_db.c
functions and stress the RegXXX() rpc functions.
(This used to be commit 0d6352da4800aabc04dfd7c65a6afe6af7cd2d4b)
|
|
to a thin layer in fetch_reg_values(). Not entirely efficient
seeing as the the dynamic value paths are stored in an unsorted
array but it is one strequal() per path. If this was really big
it should be worked into the reghook_cache().
(This used to be commit 63b81ad3cb484090a181fbd13e04922a5c17e7d9)
|
|
in init_registry_data()
* Add means of storing registry values in registry.tdb
* add builtin_registry_values[] array for REG_DWORD and REG_SZ
values needed during startup
* Finish up RegDeleteValue() and RegSetValue()
* Finish up regdb_store_reg_values() and regdb_fetch_reg_values()
I can now create and retrieve values using regedit.exe on Win2k.
bin/net -S rain -U% rpc registry enumerate 'hklm\software\samba'
Valuename = Version
Type = REG_SZ
Data = 3.0.20
Next is to do the virtual writes in reg_printing.c and I'll be
done with Print Migrator (yeah! finally)
(This used to be commit 3d837e58db9ded64d6b85f047012c7d487be4627)
|
|
* merge a compile warning fix from trunk to SAMBA_3_0
(This used to be commit 71eb018a05c5012fbd42ba6817aabc0797d38ba1)
|
|
* implement RegDeleteKey() for reg_db backend
(This used to be commit 91b81a23b8e2a096747e02fd9392ef590e7f0d61)
|
|
(This used to be commit a091b37d59d1e0228a9c8d4bd2a31e9bbaafde99)
|
|
* start adding write support to the Samba registry
Flesh out the server implementations of
RegCreateKey(), RegSetValue(), RegDeleteKey() and RegDeleteValue()
I can create a new key using regedit.exe now but the 'New Key #1'
key cannot be deleted yet.
(This used to be commit e188fdbef8f0ad202b0ecf3c30be2941ebe6d5b1)
|
|
RegOpenKey(); passing it off to the backend code for a given path
(This used to be commit 867fd3052bbfdd45856886999619e2ebc6552675)
|
|
* removing the testprns tool
(This used to be commit 81ffb0dbbbd244623507880c323a3c37e2b8dc4d)
|
|
(This used to be commit a71e104af84810f488f42cb0843976961e6f6ebe)
|
|
rename REG_CREATE_VALE -> REG_SET_VALUE
(This used to be commit 28d433351cf813c7fb57ebac0e0f4973c85f73e8)
|
|
allocation
functions so we can funnel through some well known functions. Should help greatly with
malloc checking.
HEAD patch to follow.
Jeremy.
(This used to be commit 620f2e608f70ba92f032720c031283d295c5c06a)
|
|
fix the confusion when we tdb_lock_bystring() but
we retrieve an entry using tdb_fetch_by_string.
It's now always tdb.*bystring()
(This used to be commit 66359531b89368939f0e8f584a45844b5f2f99e7)
|
|
(This used to be commit f8072d964f527dcb9b520ec06c3522524d47644f)
|
|
upper cased already.
However, if you created your registry tdb in the very early versions of jerry's
patch, you could find that usrmgr doesn't function. Simply delete the
registry.tdb, it will be recreated on startup.
Andrew Bartlett
(This used to be commit 17136a88c326bf338f948a67c92bb048c5a148af)
|
|
to check for uppercased strings when we store them not uppercased.
jerry, this fix is needed to make usrmgr.exe work again.
meanwhile we found out that NT_STATUS code may not be appropriate there.
In particular it seem that an NT PDC will send back 02 as error
(ERRbadfile) not 0xc000000f (NT_STATUS_NO_SUCH_FILE NT)
I think further investigation is need to understand which are aprropriate
return codes here.
(This used to be commit 2ad0e81c8da62b7e15ab3e414b5e15a94fe5de87)
|
|
* normalize all registry key strings before storing or looking
up paths in the registry tdb
* return the current buffer size for REG_INFO even when not returning
actual data
* fix a segfault report by metze on #samba-technical so that the
user/group object picker works again (was the "ProductType" key
lookup that was failing).
(This used to be commit 5640e6cdb213502d95fff33e06eaeed5ce3aeb76)
|
|
(even nonexistent ones). This gets rid of the Scheduling Agent icon.
* fix NT_STATUS return code for bad registry path (NT_STATUS_NO_SUCH_FILE)
(This used to be commit 915ee5c0ec0467fea23be8f309bcaa085c6ed9dd)
|
|
* added REG_OPEN_HKCR for supporting regedit.exe
* All data n a REGISTRY_VALUE is stored to a pointer now
* fixed REG_INFO to correctly display data when double clicking on
and entry in the registry editor
* Will now enumerate installed driver_info_3 data
* fixed numerous bugs related to pointer offsets, memory issues, etc..
in the registry routines
* added a simple caching mechanism to fetch_reg_[keys|values]_specific()
All that is left now is to enumerate PrinterData and I will have finished
what I started out to do....
(This used to be commit 419d7208e8384e4ad2c4dd328ad5e630971bc76c)
|
|
functions now works :-)
(This used to be commit c5768538f6cf6ee824bc6e105a3391bbc2ea8e46)
|
|
use a destroyed TALLOC_CTX*
(This used to be commit 432b9f8d7c20fbf3b2a0906c8a93272abbe43fb6)
|
|
registry values are now passed around in containers
(REGSUBKEY_CTR & REGVAL_CTR) which each possess a TALLOC_CTX.
* removed subkey_specific_fn() from REGISTRY_OPS. Is implemented
in the form of a wrapper
* temporarily broke the printing registry ops.
* implemented inheritence for the data_p of nodes in a SORTED_TREE
* All REGISTRY_KEY instances now store a valid REGISTRY_HOOK since
the default REGOSTRY_OPS structure is stored in the root of the
cache_tree.
* Probably some other change I forgot.... T
(This used to be commit e7b55e8f017e638342d9c8c1a9259000745a0298)
|
|
(This used to be commit a43d9788fa8823d678ee72470421b980165ec2b0)
|
