IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
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.
box with gcc4 and -O6...
Fix a bunch of C99 dereferencing type-punned pointer will break
strict-aliasing rules errors. Also added prs_int32 (not uint32...)
as it's needed in one place. Find places where prs_uint32 was being
used to marshall/unmarshall a time_t (a big no no on 64-bits).
More warning fixes to come.
Thanks to Volker for nudging me to compile like this.
Jeremy.
a list. Also not converting names from DOS CP to UNIX CP correctly. This
code doesn't quite work yet but it's a work in progress to be fixed
tomorrow (don't want to lose it).
Jeremy.
to make it generic. Remove the mb-codepage "blacklist". Alexander, please
check this fix as it reverts your blacklist changes, but I'm hoping it
fixes the problem in a more generic way for all charsets. I'm not trying
to trample on your (excellent!) work here, just make things more generic
without special cases.
Jeremy.
bytes in the mb character at a pointer. Will be useful in fixing check_path_syntax()
to not use a "blacklist". Also re-added my (C) to reply.c. I mean, really - I've
been adding code to the file for over 10 years and I recognise many of the
fuctions as mine ! :-).
Jeremy.
This is intermediate fix as discussed with Jeremy until we move check_path_syntax() to UCS2 internally where all
ambiguity is resolved. Please add other encodings into charcnv.c with such property.'
'
Don't use nstrings to hold workgroup and netbios names. The problem with them is that MB netbios
and workgroup names in unix charset (particularly utf8) may be up to 3x bigger than the name
when represented in dos charset (ie. cp932). So go back to using fstrings for these but
translate into nstrings (ie. 16 byte length values) for transport on the wire.
Jeremy.
boolean parameter that allows broken iconv conversions to work. Gets rid of the
nasty errno checks in mangle_hash2 and check_path_syntax and allows correct
return code checking.
Jeremy.
try a crap conversion instead. Next this needs to be done to the convert_alloc
function.
Actually fixes some valgrind warnings as well - cool !
Jeremy.
one place where we checked "if (src_len > 0)".
I actually would greatly prefer to switch back to int for src_len. The
type *can* be negative, which means an unsigned type is
inappropriate. There is absolutely no reason why "int" should not be
used for a parameter like this.
I didn't change back to int as we are close to a release and I wanted
a mininal change, but please don't go changing types like this in
future without very careful testing and a damn good reason.
this bug broke pull_ucs2(), I would not be surprised if it caused all
sorts of nastiness. Thanks to vl for noticing the symptoms!
- setup_logging() in smbclient to be interactive (remove the timestamps)
- Fix bad return value in pull_ucs2( needs more testing to make sure this
didn't break something else) that caused clistr_pull() to always read
the same string from the buffer (pull_usc2() could return -1 if the original
source length was given as -1)
- increment some debugging messages to avoid printing them out so often
- Fill in the 'backup' idea of a domain, if the DC didn't supply one. This
doesn't seem to occour in reality, hence why we missed the typo.
lib/charcnv.c:
lib/smbldap.c:
libads/ldap.c:
libsmb/libsmbclient.c:
printing/nt_printing.c:
- all the callers to pull_utf8_allocate() pass a char ** as the first
parammeter, so don't make them all cast it to a void **
nsswitch/winbind_util.c:
- Allow for a more 'correct' view of when usernames should be qualified
in winbindd. If we are a PDC, or have 'winbind trusted domains only',
then for the authentication returns stip the domain portion.
- Fix valgrind warning about use of free()ed name when looking up our
local domain. lp_workgroup() is maniplated inside a procedure that
uses it's former value. Instead, use the fact that our local domain is
always the first in the list.
Andrew Bartlett
(rant off :-). Inspired by work from Benjamin Riefenstahl <Benjamin.Riefenstahl@epost.de>.
Also add MacOSX/Darwin configure fixes.
Jerry - can we put this in 3.0 release ? :-).
Jeremy.
Removed calls to clobber_region when not compiling with developer as
they were hiding speed problems.
Added fast path to convert_string() when dealing with ascii -> ascii,
ucs2-le to ascii and ascii to ucs2-le with values <= 0x7F. This
gives a speedup of 22% on my nbench tests.
Next I will do this on convert_string_allocate.
Jeremy.
converted to pull/push_ascii. This will not work right at the moment for non
English codepages, but compiles - I will finish the work over the weekend.
Then nmbd should be completely codepage correct.
Jeremy.
Remove strdup_upper/strdup_lower from their old file, now that they have
been moved to charcnv.c
Note that string_replace assumes that s is a pstring. (doco change only)
Andrew Bartlett
- Treat the NMB names in the 'session request' packet as 'ASCII'. This means
that we do not get invalid multibyte from the wire, even if we truncate
in the conversion. (Otherwise we panic when we try to strupper_m it).
- Remove acnv_uxu2(), as it was duplicated by push_ucs2_allocate()
- Remove acnv_dosu2(), as it is not used.
- In push_ucs2(), with the STR_UPPER flag, do the case conversion *after*
the UCS2 conversion, when it we know that the length can't change. Also
faster, as we don't need to do another 2 UCS2 conversions.
Andrew Bartlett
