Full_Name: Jay Fenlason Version: 2.3.17 OS: Linux URL: Submission from: (NULL) (66.187.230.200) I was going through our bugzilla closing old OpenLDAP bugs when I found https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=160138 To my suprise, that code is still in 2.3.17, and is clearly broken. "newp" is not set in any way before the dn_match call. I would guess that the match is supposed to be against "nnewp", but I don't understand the code enough to be sure.
changed notes changed state Open to Test moved from Incoming to Software Bugs
> I was going through our bugzilla closing old OpenLDAP bugs when I found > https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=160138 > To my suprise, that code is still in 2.3.17, and is clearly broken. "newp" is > not set in any way before the dn_match call. I would guess that the match is > supposed to be against "nnewp", but I don't understand the code enough to be > sure. Correct, good catch. Yes, the test was clearly supposed to be for nnewp, which contains the normalized parent's DN, so the comparison reduces to a memcmp(); then the pretty form in newp is used to propagate the operation. Fixed in HEAD; thanks, p.
On Wed, Jan 11, 2006 at 08:33:31PM +0000, Pierangelo Masarati wrote: > > I was going through our bugzilla closing old OpenLDAP bugs when I found > > https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=160138 > > To my suprise, that code is still in 2.3.17, and is clearly broken. "newp" > is > > not set in any way before the dn_match call. I would guess that the match is > > supposed to be against "nnewp", but I don't understand the code enough to be > > sure. > > Correct, good catch. Yes, the test was clearly supposed to be for nnewp, which > contains the normalized parent's DN, so the comparison reduces to a memcmp(); > then the pretty form in newp is used to propagate the operation. > > Fixed in HEAD; thanks, p. I'm looking through all the warning messages generated when I build 2.3.17. Expect another bug report when I've gone through them all. Most of them are "warning: too many arguments for format" because of the Debug macro (I think), but a few of them look like they might be real. -- JF
On Wed, 2006-01-11 at 21:07 +0000, fenlason@redhat.com wrote: > I'm looking through all the warning messages generated when I build > 2.3.17. Expect another bug report when I've gone through them all. > Most of them are "warning: too many arguments for format" because of > the Debug macro (I think), Right, you may (safely?) ignore them; they will eventually disappear as logging gets redesigned. > but a few of them look like they might be > real. In my builds, apart from few mostly useless modules all warnings are "too many args" or "type punned/strict aliasing" stuff (gcc 3.4.4 -Wall -Wextra -pedantic). If you find anything else, please report. Thanks, p. Ing. Pierangelo Masarati Responsabile Open Solution OpenLDAP Core Team SysNet s.n.c. Via Dossi, 8 - 27100 Pavia - ITALIA http://www.sys-net.it ------------------------------------------ Office: +39.02.23998309 Mobile: +39.333.4963172 Email: pierangelo.masarati@sys-net.it ------------------------------------------
On Thu, Jan 12, 2006 at 12:24:12AM +0100, Pierangelo Masarati wrote: > On Wed, 2006-01-11 at 21:07 +0000, fenlason@redhat.com wrote: > > > I'm looking through all the warning messages generated when I build > > 2.3.17. Expect another bug report when I've gone through them all. > > Most of them are "warning: too many arguments for format" because of > > the Debug macro (I think), > > Right, you may (safely?) ignore them; they will eventually disappear as > logging gets redesigned. > > > but a few of them look like they might be > > real. > > In my builds, apart from few mostly useless modules all warnings are > "too many args" or "type punned/strict aliasing" stuff (gcc 3.4.4 -Wall > -Wextra -pedantic). > > If you find anything else, please report. Thanks, p. All the real used-before-set errors are in the bdb part of the build, so not your problem. I did notice: ../../../libraries/libldap/tls.c: In function 'tls_get_cert': ../../../libraries/libldap/tls.c:879: warning: implicit declaration of function 'ssl3_send_alert' This looks like it's using an openssl internal function, which is probably a no-no. ../../../libraries/libldap/test.c: In function 'get_modlist': ../../../libraries/libldap/test.c:172: warning: 'tmp.mod_op' may be used uninitialized in this function This looks like it could be real, but it looks like test code. The easy solution would be to bzero (erm memset) the structure before the loop. ../../../libraries/libldap_r/thr_posix.c: In function 'ldap_pvt_thread_set_concurrency': ../../../libraries/libldap_r/thr_posix.c:58: warning: implicit declaration of function 'pthread_setconcurrency' ../../../libraries/libldap_r/thr_posix.c: In function 'ldap_pvt_thread_get_concurrency': ../../../libraries/libldap_r/thr_posix.c:72: warning: implicit declaration of function 'pthread_getconcurrency' ../../../libraries/libldap_r/thr_posix.c: In function 'ldap_pvt_thread_kill': ../../../libraries/libldap_r/thr_posix.c:174: warning: implicit declaration of function 'pthread_kill' I think these are actually "bugs" in the pthread header files. I can't find a declaration for pthread_kill() anywhere, although it has a man page. The others don't have man pages, and the declarations are hidden behind an #ifdef. -- JF
Please report this separately from the syncrepl issue. At 09:03 AM 1/12/2006, fenlason@redhat.com wrote: >On Thu, Jan 12, 2006 at 12:24:12AM +0100, Pierangelo Masarati wrote: >> On Wed, 2006-01-11 at 21:07 +0000, fenlason@redhat.com wrote: >> >> > I'm looking through all the warning messages generated when I build >> > 2.3.17. Expect another bug report when I've gone through them all. >> > Most of them are "warning: too many arguments for format" because of >> > the Debug macro (I think), >> >> Right, you may (safely?) ignore them; they will eventually disappear as >> logging gets redesigned. >> >> > but a few of them look like they might be >> > real. >> >> In my builds, apart from few mostly useless modules all warnings are >> "too many args" or "type punned/strict aliasing" stuff (gcc 3.4.4 -Wall >> -Wextra -pedantic). >> >> If you find anything else, please report. Thanks, p. > >All the real used-before-set errors are in the bdb part of the build, so >not your problem. I did notice: > >../../../libraries/libldap/tls.c: In function 'tls_get_cert': >../../../libraries/libldap/tls.c:879: warning: implicit declaration of function 'ssl3_send_alert' > >This looks like it's using an openssl internal function, which is >probably a no-no. > >../../../libraries/libldap/test.c: In function 'get_modlist': >../../../libraries/libldap/test.c:172: warning: 'tmp.mod_op' may be used uninitialized in this function > >This looks like it could be real, but it looks like test code. The >easy solution would be to bzero (erm memset) the structure before the >loop. > >../../../libraries/libldap_r/thr_posix.c: In function 'ldap_pvt_thread_set_concurrency': >../../../libraries/libldap_r/thr_posix.c:58: warning: implicit declaration of function 'pthread_setconcurrency' >../../../libraries/libldap_r/thr_posix.c: In function 'ldap_pvt_thread_get_concurrency': >../../../libraries/libldap_r/thr_posix.c:72: warning: implicit declaration of function 'pthread_getconcurrency' >../../../libraries/libldap_r/thr_posix.c: In function 'ldap_pvt_thread_kill': >../../../libraries/libldap_r/thr_posix.c:174: warning: implicit declaration of function 'pthread_kill' > >I think these are actually "bugs" in the pthread header files. I >can't find a declaration for pthread_kill() anywhere, although it has >a man page. The others don't have man pages, and the declarations are >hidden behind an #ifdef. > > -- JF
On Thu, 2006-01-12 at 12:03 -0500, Jay Fenlason wrote: > All the real used-before-set errors are in the bdb part of the build, so > not your problem. I did notice: > > ../../../libraries/libldap/tls.c: In function 'tls_get_cert': > ../../../libraries/libldap/tls.c:879: warning: implicit declaration of function 'ssl3_send_alert' > > This looks like it's using an openssl internal function, which is > probably a no-no. ^^^ this has already been discussed in the past; I do not recall the exact reason (and I'm too lazy to look it up in the archives) but yes, it's calling an internal for something that has to be done that way. Call it a "limitation" of ssl API... > > ../../../libraries/libldap/test.c: In function 'get_modlist': > ../../../libraries/libldap/test.c:172: warning: 'tmp.mod_op' may be used uninitialized in this function > > This looks like it could be real, but it looks like test code. The > easy solution would be to bzero (erm memset) the structure before the > loop. right; but ll in all it's a harmless test. > > ../../../libraries/libldap_r/thr_posix.c: In function 'ldap_pvt_thread_set_concurrency': > ../../../libraries/libldap_r/thr_posix.c:58: warning: implicit declaration of function 'pthread_setconcurrency' > ../../../libraries/libldap_r/thr_posix.c: In function 'ldap_pvt_thread_get_concurrency': > ../../../libraries/libldap_r/thr_posix.c:72: warning: implicit declaration of function 'pthread_getconcurrency' > ../../../libraries/libldap_r/thr_posix.c: In function 'ldap_pvt_thread_kill': > ../../../libraries/libldap_r/thr_posix.c:174: warning: implicit declaration of function 'pthread_kill' > > I think these are actually "bugs" in the pthread header files. I > can't find a declaration for pthread_kill() anywhere, although it has > a man page. The others don't have man pages, and the declarations are > hidden behind an #ifdef. Right. Thanks, p. Ing. Pierangelo Masarati Responsabile Open Solution OpenLDAP Core Team SysNet s.n.c. Via Dossi, 8 - 27100 Pavia - ITALIA http://www.sys-net.it ------------------------------------------ Office: +39.02.23998309 Mobile: +39.333.4963172 Email: pierangelo.masarati@sys-net.it ------------------------------------------
changed notes changed state Test to Release
changed state Release to Closed
moved from Software Bugs to Archive.Software Bugs
fixed in HEAD/re23