Hi Irek,

V po, 06. 07. 2009 v 23:18, I. Szczesniak p??e:
> Roland, Milan,
> http://svn.genunix.org/repos/on/branches/ksh93/gisburn/prototype021/usr/src/pkgdefs/SUNWcsu/prototype_com
> has cksum listed as file, not as hard link:
> 
> f none usr/bin/cksum 555 root bin
> 
> The warning will disappear if you change this line to
> 
> l none usr/bin/cksum=../../usr/bin/alias
> 
> i probably should've caught it during earlier code reviews. Sorry.
> 

Excellent catch :-) I was looking at it several times but I missed it.

Best regards,

Milan

> Irek
> 
> On 7/6/09, Milan Jurik <Milan.Jurik at sun.com> wrote:
> > Hi Roland,
> >
> >  CCed ksh93-integration-discuss@
> >
> >  Roland Mainz p??e v po 06. 07. 2009 v 21:14 +0200:
> >  > Milan Jurik wrote:
> >  > > so the last thing I must discover it is the problem with ref_cnt
> >  > > discrepancy between proto area and pkgdefs.
> >  >
> >  > I am not seeing these differences... how do you enable these tests ?
> >  >
> >
> >  nightly options "-aADClmprfzt" as I wrote you several days ago.
> >
> >  > > Then I will push LU and VSC
> >  > > testing the next week.
> >  >
> >  > Can you please run the VSC tests in parallel, please ? AFAIK they are
> >  > independent from the filesystem link count issues...
> >  >
> >
> >  Probably yes. I think we will start with LU and VSC tomorrow.
> >
> >  > [snip]
> >  > > Files that changed between yesterday and today:
> >  > >
> >  > > Unit   T File Name                      Reloc/Sym name       perm 
> > owner group inode lnk maj min package(s)
> >  > > 
> > ------------------------------------------------------------------------------------------------------------
> >  > > filea: f usr/bin/cd                     -                     555 -    
> >  -     341954 20  -  -    proto
> >  > > fileb: f usr/bin/cd                     -                     555 -    
> >  -     2718295 35  -  -    proto
> >  > >     differ: ref_cnt
> >  >
> >  > Erm... isn't this only the file link refernce count which changed here ?
> >  > The alias.c wrapper now "manages" more commands and therefore the link
> >  > count increases...
> >  >
> >
> >  I agree and it is not the problem.
> >
> >  > [snip]
> >  > > Inconsistencies between pkgdefs and proto area:
> >  > >
> >  > > Unit   T File Name                      Reloc/Sym name       perm 
> > owner group inode lnk maj min package(s)
> >  > > 
> > ------------------------------------------------------------------------------------------------------------
> >  > > filea: l usr/bin/cd                     ../../usr/bin/alias   555 -    
> >  -          0 34  -  -    SUNWcsu
> >  > > fileb: f usr/bin/cd                     -                     555 -    
> >  -     2718295 35  -  -    proto
> >  > >     differ: ref_cnt
> >  > > filea: l usr/bin/bg                     ../../usr/bin/alias   555 -    
> >  -          0 34  -  -    SUNWcsu
> >  > > fileb: f usr/bin/bg                     -                     555 -    
> >  -     2718295 35  -  -    proto
> >  >
> >  > Mhhh... weired...
> >  > ... is it possible that this is caused by having /usr/xpg4/bin/tail in a
> >  > different package than "SUNWcsu" ?
> >  >
> >
> >  Yes. I double checked proto area and there was usr/bin/alias and 34
> >  links to it representing the commands. So link count 35. Additionally I
> >  tried:
> >
> >  1) onnv nightly without changes - no problem then
> >  2) usr/xpg4/bin/tail in SUNWcsu - the problem remains
> >
> >
> >  Best regards,
> >
> >  Milan
> >
> >
> >  > ----
> >  >
> >  > Bye,
> >  > Roland
> >  >
> >  > P.S.: Can I include ksh93-integration-discuss at opensolaris.org to this
> >  > reply, please ?
> >  >
> >
> >  _______________________________________________
> >  ksh93-integration-discuss mailing list
> >  ksh93-integration-discuss at opensolaris.org
> >  http://mail.opensolaris.org/mailman/listinfo/ksh93-integration-discuss
> >
> _______________________________________________
> ksh93-integration-discuss mailing list
> ksh93-integration-discuss at opensolaris.org
> http://mail.opensolaris.org/mailman/listinfo/ksh93-integration-discuss


Reply via email to