Code went into the cvs last night.  Please test it.

-----Burton


> -----Original Message-----
> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf
> Of Burton M. Strauss III
> Sent: Monday, January 26, 2004 6:40 AM
> To: [EMAIL PROTECTED]
> Subject: RE: Re[4]: [Ntop-dev] gzflush problems
> 
> 
> You would have to download the matching kernel source package and examine
> the .config file they're actually using.  The problem is that RH 
> backports a
> lot of patches, so just knowing the base kernel version isn't helpful.
> 
> I really don't know why it fixes things - all I can guess is that you're
> using the kernel version instead of the libz.so version and the two are
> different in some way.
> 
> Regardless, the right answer is to make the MAKE_WITH_ZLIB fix in 
> the code,
> so if you request it be disabled @ ./configure time, it is...
> 
> -----Burton
> 
> > -----Original Message-----
> > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf
> > Of Vilius Sumskas
> > Sent: Sunday, January 25, 2004 1:21 PM
> > To: Burton M. Strauss III
> > Subject: Re[4]: [Ntop-dev] gzflush problems
> >
> >
> > Hello Burton,
> >
> > Sunday, January 25, 2004, 8:41:29 PM, you wrote:
> >
> > > Interesting.  And it makes NO SENSE.
> >
> > > Because all that function really does is suppress the TESTING
> > for zlib in
> > > ./configure.  There's no code changes that respect the
> > MAKE_WITH_ZLIB flag.
> > > All the code that calls gzopen() et al is still in there and
> > still live...
> >
> > > Wild guess.  Is this Linux?  New Linux, like 2.4.20+ with the
> > zlib kernel
> > > option???
> >
> > > -----Burton
> >
> >
> > >> -----Original Message-----
> > >> From: [EMAIL PROTECTED]
> > >> [mailto:[EMAIL PROTECTED] Behalf
> > >> Of Vilius Sumskas
> > >> Sent: Sunday, January 25, 2004 12:14 PM
> > >> To: Burton M. Strauss III
> > >> Subject: Re[2]: [Ntop-dev] gzflush problems
> > >>
> > >>
> > >> Hello Burton,
> > >>
> > >> Sunday, January 25, 2004, 5:55:35 PM, you wrote:
> > >>
> > >> > It's not MAKE_WITH_ZLIB 0, it's #undef MAKE_WITH_ZLIB
> > >>
> > >> > #define MAKE_WITH_ZLIB 0
> > >>
> > >> > means that the tests, e.g #ifdef MAKE_WITH_ZLIB, will 
> succeed (it IS
> > >> > defined, right?)
> > >>
> > >> > Now the underlying problem is one of malloc() memory
> > >> corruption, which has
> > >> > not occurred in any current version of the OS/glibc that I'm
> > >> aware of since
> > >> > late summer.  You may need to upgrade your glibc (which is a huge
> > >> > undertaking - better to update the whole OS, as it's less work).
> > >>
> > >> > I'll look at putting --without-zlib back in to
> > >> > ./configure, but if it causes
> > >> > problems in the code, I will NOT do it.
> > >>
> > >>
> > >> Just checkedout cvs. Configured and compiled with 
> --without-zlib. Both
> > >> gzflush error AND problem with .gz files fixed. Thanks a lot!
> > >>
> > >>
> > >> --
> > >> Best regards,
> > >>  Vilius
> >
> > It is redhat 7.2 with 2.4.20-28.7 kernel. How do I check if it is with
> > "zlib kernel option" ?
> >
> > --
> > Best regards,
> >  Vilius
> >
> > _______________________________________________
> > Ntop-dev mailing list
> > [EMAIL PROTECTED]
> > http://listgateway.unipi.it/mailman/listinfo/ntop-dev
> >
> 
> _______________________________________________
> Ntop-dev mailing list
> [EMAIL PROTECTED]
> http://listgateway.unipi.it/mailman/listinfo/ntop-dev
> 
_______________________________________________
Ntop-dev mailing list
[EMAIL PROTECTED]
http://listgateway.unipi.it/mailman/listinfo/ntop-dev

Reply via email to