[EMAIL PROTECTED] writes:
Modified Files:
slapi_overlay.c 1.28 - 1.29
pb in slapi_op_func was freed uninitialized if preop_type==0 !internal_op
Don't know if I fixed that in the right place. I NULLed out pb before
doing goto cleanup:
cleanup:
if ( !internal_op ) {
[EMAIL PROTECTED] writes:
init.c 1.111 - 1.112
more (in)sanity stuff
I haven't looked closely, but I imagine cleaner fix would be to move
parts of monitor_back_db_destroy (bi-bi_db_destroy) to bi-bi_destroy.
Or maybe parts of monitor_back_initialize to monitor_back_db_init
Kurt D. Zeilenga writes:
At 01:58 AM 8/24/2005, Howard Chu wrote:
Anyway, after fixing the OL_PTHREAD_TRY macro it appears that this
configure script runs OK on my SuSE 9.2 x86_64 system. make test is
running now, looks OK.
./configure --help shows that there is a problem with
The description of '-q' in slapadd(5) says:
enable quick (fewer integrity checks) mode. Does fewer consistency
checks on the input data, and no consistency checks when writing the
database. Improves the load time but if any errors or interruptions
occur the resulting database will
Quanah Gibson-Mount writes:
[EMAIL PROTECTED] wrote:
slapadd -q has been entirely safe for me as long. The point of that
paragraph is that if you kill slapadd, or an error occurs during the
slapadd, you will need to wipe the database and start over. Apart from
that, it is completely safe. ;)
Howard Chu writes:
Though actually I would like an option for more LDIF error checking
too: Adding whatever check ldapadd does which is convenient to add to
slapadd, including a verify function in overlays like unique, to
be called for each entry after the database is built and indexed.
tests/data/slapd-valsort.conf is missing, for test042.
tests/data/slapd-aci.conf is missing.
If it is copied from HEAD, slapd dumps core in test041:
#0 0x080a807b in parse_acl (be=0x82b75a8,
fname=0x828b008 ./testrun/slapd.1.conf,
lineno=54, argc=6, argv=0x82b48f0, pos=-1) at
I'll add support for slapadd, slapcat etc to the null backend.
I actually found a use for it recently:-)
--
Hallvard
Backends can have options with the same name but very different
semantics, and so far this seems to be reflected in back-config
attribute names. That could get messy, in particular with contrib or
private backend modules written using some other backend as a template.
I think a config
slapd is often too silent when something fails.
One may need -d -1 or something to find out what is wrong -
and then useful messages can drown in others.
Also, if OpenLDAP is configured with --disable-debug, often
we get no error messages at all. That's plain wrong - fatal
error messages are not
How do I build configure in HEAD now? autoconf-2.59 + this patch?
http://lists.gnu.org/archive/html/autoconf/2005-08/msg00076.html
Subject:Re: [Fwd: libtool AC_ARG_WITH and autoconf 2.59]
From: Howard Chu [EMAIL PROTECTED]
To: autoconf@gnu.org
Date: Thu, 25 Aug 2005
I wrote:
I've been sitting on a thread debugging package for a while now, and it
seems to have stabilized. I'll commit it soon unless someone thinks it
should be done differently.
...committed to HEAD.
--
Hallvard
Which uses of AVLs need optimzation?
A better speedup in some cases should be to replace them with hash
tables, splay trees, skip lists or something.
Maybe there is a package out there which has the same interface to
several such data structures, so we easily can plug in different ones
for
[EMAIL PROTECTED] writes:
Modified Files:
run.in 1.40 - 1.41
Source scripts/defines.sh _after_ the variables it uses have been set
Erm, that referred to some private variables of mine:-)
Oh well, the change doesn't hurt.
--
Hallvard
Pierangelo Masarati writes:
In any case this is mainly a hack to allow importing the project
under AEGIS; I don't either like having to source defines.sh twice
only to get few directories. Maybe directories should simply be
set up in run and passed to defines.sh thru the environment?
I
Pierangelo Masarati writes:
Hallvard B Furuseth writes:
* Factor a lot of code out to shell functions in defines.sh. (...)
Shell functions are fairly portable now, and in any case OpenLDAP
uses build/shtool which uses shell functions. (...)
You seem to be fairly optimistic about
Walter Franzini writes:
Hallvard B Furuseth wrote:
Pierangelo Masarati writes:
I suppose ./run could run the scripts by sourcing them instead:
test -x $SCRIPT (. $SCRIPT)
Then the clean operations at the end of ./run will never be executed.
No, the () puts the script in a subprocess
Pierangelo Masarati writes:
test040 is currently specific to back-hdb because only back-hdb
supports subtree rename (it could be moved to a backend-specific
add-on to test005)
That's true. Though if back-newWonderfulDatabase is implemented someday
and also supports subtree rename, test005
Howard Chu writes:
I guess it would be nice to fix things so we don't have to edit the
scripts and mess up our commits, but how do you set up the variables
in a script that invokes slapd (or any program) multiple times, and
you only want special treatment for one specific invocation?
Well...
[EMAIL PROTECTED] writes:
Modified Files:
defines.sh 1.129 - 1.130
allow writable dirs to be user-relocated
-DATADIR=./testdata
PROGDIR=./progs
-TESTDIR=./testrun
-SCHEMADIR=./schema
+DATADIR=${USER_DATADIR-./testdata}
+TESTDIR=${USER_TESTDIR-./testrun}
I'm trying to maintain OpenLDAP software using Aegis. One of the
constraints imposed by Aegis is that the integration directory can not
be modified by the testsuite.
Looking at the various scripts under tests/ it seems to me that the
solution for my problem is to set TESTDIR=/tmp/.
[EMAIL PROTECTED] writes:
test007-replication 1.67 - 1.68
test018-syncreplication-persist 1.35 - 1.36
ldapmodify doesn't return an LDAP error code
How did we miss that for so long?
The ldapmodify source code looks like it intends to return
a result code, though. And it does
Pierangelo Masarati writes:
To: Hallvard B Furuseth [EMAIL PROTECTED]
Maybe we should change all the LDAP tools to always return some
LDAP error code, and some special code when it cannot or when an
LDAP code does not make sense?
The point is: LDAP error codes are integers, while command
Howard Chu writes:
On the other hand, when exactly is it possible for a single request to
have more than one result message?
SearchResultEntry/Reference and the newer IntermediateResponse. I
seem to remember there was an extended search result or something once
which was removed due to lack of
Kurt D. Zeilenga writes:
What about the case where ldap_result(3) is called for any and
all? The message chained returned could contain responses
from multiple requests.
ldap_result() should return responses to one specific request.
ldap_result(,msgid=LDAP_RES_ANY,,) may pick any outstanding
Howard Chu writes:
It certainly sounds like an improvement over back-shell, and there may
be some value in keeping the perl interpreter outside the slapd process.
Yes: And slapd won't crash if a Perl routine crashes.
And perl can be run as a user with none of slapd's privileges, so
one need not
The terminal escape sequences from 'make test outfile' can be a bit
annoying. It's not hard to say 'TERM=dumb make test outfile', but
I forget it now and then.
I'd like to wrap the code to find terminal escape sequences in configure
and 'make test' in 'if test -t 1', which checks if file
Howard Chu writes:
OK, suppose we add a -g option that disables glue completely?
Fine. Or possibly -G, since I have a vague feeling that uppercase
is more common for options that negate what they are named for.
--
Hallvard
Howard Chu writes:
This field is gratuitously set to LBER_INITIALIZED throughout some
heavily used liblber routines. There appears to be only one place
where the value is ever checked (in ber_set_option).
What will you replace it with?
Its function seems to be (slightly simplified):
When a
Pierangelo Masarati writes:
I fear yet more few minor concurrency issues in libldap; I'd like to
be able to assert if a mutex is locked from inside library functions,
and it would be great to also know who's locking it. What I'd do is
#define ISLOCKED(m) \
( ldap_pvt_mutex_trylock(
One thing which is missing is pthread_once() to protect init functions.
I'll commit one for pthreads and a default version, unless someone
protests.
Then e.g. ldap_int_initialize() can use that - and if two threads call
it at the same time, one will wait for the other's initialization to
Pierangelo Masarati writes:
Well, what I really need as a baseline is if some mutex is not locked
at all. If it's not, then it's an error. If it is, it's an error
only if it's not owned by the caller.
Can you give an example of some code which needs this?
I can't offhand think of an problem
Pierangelo Masarati writes:
I fear yet more few minor concurrency issues in libldap; I'd like to be
able to assert if a mutex is locked from inside library functions,
Duh... please ignore my previous message. I forgot you said _assert_.
Well, that one isn't hard.
define an
Pierangelo Masarati writes:
Playing with Hallvard's thread debugging code I came into something
that looks strange: slapd does tpool init/cleanup fine, while tools
don't.
(...)
We discussed a related issue some time ago. The whole init/cleanup
stuff in slapd is skewed. main() does
Pierangelo Masarati writes:
On Fri, 2006-03-31 at 20:09 +0200, Hallvard B Furuseth wrote:
Or more generally, use my thr_debug.c package and extend
ldap_debug_thread_cond_t with the thread ID and info about whether the
mutex is locked. Get thread ID with ldap_pvt_thread_self(), compare
Pierangelo Masarati writes:
Another issue that's happening right now is that ldap_pvt_thread_exit()
may be called even if no ldap_pvt_thread_pool_submit() was issued (e.g.
try starting slapd and killing it before any operation is run); this
causes the thread debug build to fail with
slapd
Any reason why test008-concurrency uses loglevel -1 instead of
$LVL (${SLAPD_DEBUG-261}) from defines.sh?
A little excessive when for hosts with little disk space...
--
Hallvard
[EMAIL PROTECTED] writes:
result.c 1.292 - 1.293
re-fix ITS#4495 working around atoi() limitations
Wow, that was a lot of debug messages:-)
Still needs work though, e.g. strip whitespace from all input lines.
output code: $RC
output info:urgle
And I imagine that we could
Howard Chu writes:
I now see this message when slapd exits:
== thr_debug: Leaked 2 unexited threads, 46 mutexes. ==
Of course, running with tracethreads shows that slapd created and
joined only 2 threads, so there isn't really any thread leak. This
message doesn't seem to mean anything then.
[EMAIL PROTECTED] writes:
threads.c 1.19 - 1.20
rmutex_lock etc.: caller supplies thread id
Why is the kludge to pull symbol definitions in necessary -
and if it's necessary there, why doesn't the library need to be
full of similar kludges for other symbol definitions?
--
Howard Chu writes:
This is just a quickie to get the functionality working.
It doesn't though:-)
Opheader.oh_tid (alias Operation.o_tid) is never set, yet
overlays/accesslog.c uses the value as the owner thread ID.
Did it exist once and get replaced with the 'ctx' things, or
something like
Howard Chu writes:
Hallvard B Furuseth wrote:
Opheader.oh_tid (alias Operation.o_tid) is never set, yet
overlays/accesslog.c uses the value as the owner thread ID.
(...)
Hm, I don't know what happened with oh_tid, but I think it would be
better to actually set it, in case others need
Luke Howard writes:
Yes, memory debuggers like Valgrind can show a stack trace of where the
resource was created. thr_debug can't do that, so the output is more
informative when it provokes Valgrind instead.
On Linux you can call backtrace_symbols_fd() to get a backtrace from
within your
[Pierangelo Masarati]
What about a -n (much like sophisticated echo implementations)
to avoid breaking existing usage?
Actually it's not simple vs sophisticated, but the SysV echo foo\c
vs. BSD echo -n foo. Except today we've got such mixtures of SysV +
BSD that it is hard to tell.
[Aaron
Just a loose idea...
Ownership often seems a bit hairy to keep track of, like ITS#4550
just now. I've been wondering if it would ease things up to add
reference counts to some objects, but I don't know just when it
would be useful...
Maybe berval-like constant strings instead of
man slapo-ppolicy, option ppolicy_hash_cleartext, says:
It is recommended that when this option is used that compare,
search, and read access be denied to all directory users.
Huh?
It would make sense to me if it said ...when this option is *not*
used ... read access *to userPassword*
Attribute (OLcfgDbAt:0.1 NAME 'olcDbDirectory') has different
definitions in back-ldif and back-bdb: No SINGLE-VALUE in back-ldif.
Looks like a likely bug to reappear when other common attributes are
added. I suggest that OIDs in the common namespace are defined in
bconfig.c (in this case with
I wrote:
In any case, it's a nice feature that e.g. phone numbers in an entry can
be returned in the preferred order. So if it doesn't get cumbersome, I
would prefer to be able to turn a don't reorder feature on in
slapd.conf.
I mean, turn it on for individual attributes.
--
Hallvard
Kurt D. Zeilenga writes:
At 12:09 PM 8/16/2006, Howard Chu wrote:
This issue has been around for a long time, and I think we're going
to need to deal with it soon, as we see sites creating groups with
thousands of members.
Yup. Our site currently has 6 with 2 members, 7 more with 5000.
I've had it with explaining that the suffix is the root of the database
which is not the rootdn which is not the Unix user root. Any hope of
renaming rootdn and rootpw to admindn and adminpw, with rootdn/pw as
backwards-compatible and eventually undocumented aliases?
--
Hallvard
Michael Ströder writes:
On 2:00:38 am 2007-01-10 Kurt D. Zeilenga [EMAIL PROTECTED] wrote:
At 11:50 AM 1/6/2007, Hallvard B Furuseth wrote:
I've had it with explaining that the suffix is the root of the
database which is not the rootdn which is not the Unix user root.
Any hope of renaming
Pierangelo Masarati wrote:
Howard Chu wrote:
On a side-note, when I was benchmarking the connection manager I
found it useful to doctor up back-null to actually return an entry in
response to search requests. If anyone else would find this useful, I
can commit that patch.
For a similar need,
Howard Chu writes:
Pierangelo Masarati wrote:
[EMAIL PROTECTED] wrote:
Update of /repo/OpenLDAP/pkg/ldap/libraries/libldap
Modified Files:
result.c 1.149 - 1.150
Log Message:
Fairly sure this is what the TIMEOUT option should always have been for
Not strictly related, but does it
Howard Chu writes:
Hallvard B Furuseth wrote:
Do you mean ldap_get_option()? It generally returns malloced data,
IIRC. undocumented just means the code has been the documentation,
so I presume a change will break any old code which uses it.
Yes to all of the above. But since
Howard Chu writes:
The documentation for this feature presents a bit of a problem, since
most of the functionality of librewrite is documented in
slapo-rwm(5). When SLAP_AUTH_REWRITE is defined (which it is, whenever
--enable-rewrite is used) then all of librewrite's capabilities really
Howard Chu writes:
It's almost aggravating enough to make me switch to pure DAP and leave
all of this cruft behind. Mebbe it's time to add a config switch
somewhere to set all of slapd and the libraries to default to
DCE/X.500 format DNs, and define a new xdap URL scheme to go with
it.
DNs
As far as I can tell, OpenLDAP has been unusable with KR C since
openldap-2.0: lber.h in that release had prototypes unprotected
by LDAP_P() in struct sockbuf_io. They've been there ever since.
So LDAP_P() can go.
Don't know which other not-strict-ISO support is still useful -
I note lber.h has
Kurt Zeilenga writes:
On Jun 14, 2007, at 9:44 AM, Hallvard B Furuseth wrote:
As far as I can tell, OpenLDAP has been unusable with KR C since
openldap-2.0: lber.h in that release had prototypes unprotected
by LDAP_P() in struct sockbuf_io. They've been there ever since.
So LDAP_P() can go
[EMAIL PROTECTED] writes:
For ITS#4995, add clarification comments
How about
perlModulePath /directory/with/this/module
?
--
Regards,
Hallvard
Gavin Henry writes:
I've run Perl::Critic over this and fixed/updated to latest Perl Best
Practices. We might as well have an up to date sample.
Do you mind if I commit my changes?
My impression is that back-perl hovers on the verge of code rot, and any
change is likely for the better.
As
Pierangelo Masarati writes:
doc/man/man5/slapd-ldbm.5 (obviously ;-) )
we could leave that file in place, just stating that it's no longer
available and that slapd-bdb(5) slapd-hdb(5) is the suggested replacement.
Sounds good.
--
Regards,
Hallvard
Luca Scamoni writes:
I'm seeing failure with current HEAD. Anyone else?
Yes. ITS#4839 http://www.openldap.org/its/?findid=4839.
--
Regards,
Hallvard
Dagobert Michelsen writes:
Am 19.06.2007 um 17:22 schrieb Hallvard B Furuseth:
Is there a max number of Perl interpreters per backend?
I suppose you mean per database.
Right.
There is exactly one interpreter in use for each active connection
and each configured back-perl database because
I wrote:
Gavin Henry writes:
I've run Perl::Critic over this and fixed/updated to latest Perl Best
Practices. We might as well have an up to date sample.
Do you mind if I commit my changes?
My impression is that back-perl hovers on the verge of code rot, and any
change is likely for the
Gavin Henry writes:
Nope, you are right. My mistake. It does work, but should be:
print {*STDERR} Here in new\n;
(...)
See:
http://search.cpan.org/~thaljef/Perl-Critic-1.06/lib/Perl/Critic/Policy/InputOutput/RequireBracedFileHandleWithPrint.pm
Heh. Capital-letter STDERR stands out enough
Gavin Henry writes:
I've been trying to follow defines.sh to find a few variables. Is there
any reason the individual sections aren't alphabetised?
Some variables cannot be sorted (e.g. DBDIR1 is set from DBDIR1A),
and some variables are grouped logically within a section (e.g.
P*SRSLAVECONF
#define LDAP_MEMORY_DEBUG and SLAP_NO_SL_MALLOC breaks slapd in a
number of places. Some of the fixes are straightforward, but others
require more sweeping changes - or leaving LDAP_MEMORY_DEBUG useless,
which it has been for some time anyway.
However the LDAP_MEMORY_DEBUG troubles match a
oops, typos:
it out. (First stage: /free/ch_free/ in slapd, #define CH_FREE in all
sorry, s/free/ch_free/.
Since it can be a bit hard to keep track of which memory comes from
where, ch_free() fetches the current thread's context if passed a
NULL context.
It fetches the current
Pierangelo Masarati writes:
AFAIK, the attribute and so is recognized, but it's not implemented
(nor won't, as it is no longer needed).
If it's no longer needed - what has changed? I thought it was invented
because the existing scheme of '{hash method}' in userPassword broke the
LDAP standard.
Bug reports with huge logs but without loglevel 256 can be amazingly
useless... I'll commit this change to slapd.conf(5):
@@ -559,3 +559,3 @@
.B (0x100 stats)
-stats log connections/operations/results
+connections, LDAP operations, results (recommended)
.TP
@@ -615,3 +616,6 @@
level is
[EMAIL PROTECTED] writes:
Tag: OPENLDAP_REL_ENG_2_3
entry.c 1.129.2.13 - 1.129.2.14
import fix to ITS#5071
This (/* require ';binary' when appropriate (ITS#5071) */) is a
functionality change which can prevent people from upgrading.
I don't think that belongs so late in RE23's
Pierangelo Masarati writes:
This (/* require ';binary' when appropriate (ITS#5071) */) is a
functionality change which can prevent people from upgrading.
I don't think that belongs so late in RE23's life cycle.
You can't load a certificate without ';binary' using ldapadd/ldapmodify;
this
Gavin Henry writes:
What's the best practice for choosing the numbers in OLcfgOvAt:14.1
Above is taken from translucent.c
Not sure what you mean. OLcfgOvAt:14.1 is an attribute type defined in
translucent.c, you shouldn't take OIDs below it.
If you want to add another config attribute to
Gavin Henry writes:
Do any of you use indent or similar for cleaning up OpenLDAP code and
have a ~/.my_rules to share?
I don't know how to make indent output OpenLDAP's spacing around parens.
And I'm not sure what the rules are anyway. Space inside but not
outside parens in function calls,
Howard Chu writes:
If support for 4.2 is dropped in the next 2.4 beta, what is the earliest
version we think we will support?
We already disallow use of 4.3 due to its instability, so 4.4 would be
the minimum.
What's the best BerkeleyDB version to use with OpenLDAP 2.3 anyway?
--
Hallvard
[EMAIL PROTECTED] writes:
Modified Files:
backend.c 1.384 - 1.385
Just noticed your indentation... Note that OpenLDAP code looks
best when edited and displayed with tab-width 4, not 8.
--
Hallvard
Buchan Milne writes:
I need the patch below (since 2.3.37, maybe before) to be able to compile
backglue.c on gcc 2.9x.
(...)
- int rc;
BackendDB *b0 = op-o_bd;
op-o_bd = glue_back_select( b0, dn );
+ int rc;
That breaks on C90 compilers. Declaration after
Pierangelo Masarati writes:
I guess you need the reverse: the declaration of rc is incorrectly
placed after the invocation of glue_back_select() in current 2.3 code.
Ooh. I looked at HEAD:-)
--
Hallvard
[EMAIL PROTECTED] writes:
memberof.c NONE - 1.1
Your 'memberOf' attribute definition says
{ ( 1.2.840.113556.1.2.102 (...)
EQUALITY distinguishedNameMatch /* added */
USAGE directoryOperation /* questioned */
/*
Howard Chu writes:
This is the first stage in a larger transition, to using sorted arrays
for multivalued attributes. I haven't quite decided how the next steps
will play out just yet.
(...)
We can also implement this without any user-visible change, if we keep
an auxiliary ordering array,
Howard Chu writes:
Add a slapd.conf directive which inserts an implicit X-ORDERED 'VALUES'
in an already existing attribute definition. (...)
OK, that makes sense. Except it's not the same as X-ORDERED 'VALUES'
behavior. (...)
Yes, I guess that should have been the opposite, an
Howard Chu writes:
The Modify operation has already completed though, there's nothing to
return an error message to by then.
I see.
We do as much pre-checking as possible to validate the syntax of changes
before committing them. But here you've got an input with perfectly valid
syntax. The
Howard Chu wrote:
Hallvard B Furuseth wrote:
So let me step back a bit: I'd like slapd to send the ModifyResponse
_after_ the change has taken effect (or failed to take effect).
That's actually the way it has been implemented. Just that in this
case, we weren't validating the olcDbDirectory
Howard Chu writes:
Hallvard B Furuseth wrote:
(...) Though that's one thing which needs some documentation - how
careful should one be with cn=config changes?
That seems like a pretty silly question. cn=config is so potentially
dangerous that in OpenLDAP 2.3 we restricted it to only be usable
Howard Chu writes:
2.3.39 back-null got 27k auths/sec vs 2.4(.7)'s 36k/sec. 33% gain,
obviously back-hdb requires more CPU...
Which hash method in userPassword? I'm writing this on a host which
needs 12 seconds to do 27k crypt(password, $md5-salt$hash)-calls.
--
Hallvard
I wrote:
for larger absoulte values:
same, but for the sake of speed read the value as hex, so we only
need read the first digits (skipping initial zeros).
whoops, I meant only the first bytes of the value-as-hex will be
included in the index key.
--
Hallvard
[EMAIL PROTECTED] writes:
slapindex.8 1.28 - 1.29
Document truncate mode
Please clarify the doc of this.
I mean, I had no idea slapindex had a non-truncate mode. Does slapindex
by default not delete old indexes, or does is mean something else? When
should one _not_ use truncate mode?
Howard Chu writes:
Hallvard B Furuseth wrote:
Does slapindex by default not delete old indexes
Right. It only adds new data by default.
(...)
In BerkeleyDB, you cannot use truncate mode with transactions. So in
the default mode (without -q), back-bdb/hdb silently removes the
truncate flag
Michael Ströder writes:
It really depends on what you want to express. Note that represents
the rootDSE or root naming context. So would reference something
existing or well-defined. IMHO NULL (or None in Python) would better
signal something undefined or none-existing.
Yes. There is one
Howard Chu writes:
The -O2 build is faster from about 4 to 24 client threads. From 28 on
up, the nonoptimized code is faster at every load level. I was
originally using gcc 4.1.2 but I'm seeing the same result now using
gcc 4.2.2. Also, slapd is only configured with 8 worker threads in all
of
Still wrong. Current ordering:
indexvalue(hex value)
04 -2147483648 (-0x8000)
040001 -2147483647 (-0x7fff)
044000 -1073741824 (-0x4000)
046000 -9126805504 (-0x22000)
047000 -268435456
Finally had a look at the implementation of this (in RE24). It is
wrong. The following values produces these indexes, ordered as:
0: 048000
+1: 048001
-1: 0480ff
+257: 048101
-257: 048000feff
To get the correct ordering,
Howard Chu writes:
Hallvard B Furuseth wrote:
Got crashes due to a write after 'tmp', incremented size by 1.
Haven't looked at why yet, or whether that's the complete fix.
That should never happen. Do you have the backtrace from when the
write occurred?
No, got memory corruption later
Howard Chu writes:
Hallvard B Furuseth wrote:
I wrote:
Also the index is wrong for huge numbers. At some point the indexing
should just give up and use max/min values, but I suppose at least
cryptograpy-sized numbers should be usefully indexed. I.e. at least
two length bytes.
I wonder
RE24 passes all tests on RedHat x86_64, but should test008 spew out
'slapd-bind PID=: ldap_sasl_bind_s: Invalid credentials (49)'?
testrun/ and typescript: http://folk.uio.no/hbf/test008out.tbz.
Configured: --with-tls --with-cyrus-sasl --enable-crypt --enable-aci
--enable-unique
Howard Chu writes:
This code is wrong:
(..)
On most 32 bit machines the if( chop ...) comparison must always
fail, since ber_slen_t is only 32bits. You should do the test before
multiplying chop by 3.
No, chop already fits in ber_slen_t, since it is a ber_len_t/7*3.
The check is for overflow
Howard Chu writes:
Passing structures by value may be perfectly legal now, but we never
do it in this source base. Why start now?
We don't? OK, it's fine to do two struct copies instead inside the
function.
--
Regards,
Hallvard
Howard Chu writes:
I think you misunderstood.
No, I wrote too briefly.
No matter what you're actually testing for, there is no way you're going to
get a TRUE result from comparing ( signed 32 bit integer 0x7fff ).
Right, and chop = (max ber_len_t)/2. so the test is a no-op on
http://folk.uio.no/hbf/test008out2.tbz
ldapsearch.flt = ldif.flt + cn=James A Jones 4,ou=People,dc=example,dc=com
If you are using gcc, it might help to compile BDB and/or OpenLDAP
without -O and see if that helps. The gcc mailinglist and
comp.programming.threads recently discussed a some gcc optimizations
that are invalid for threaded code. E.g. gcc can turn things like
static or extern int v;
...
if
1 - 100 of 299 matches
Mail list logo
