alsa-sane-headers sanity check broken

Tue Nov 23 02:23:59 UTC 2010

On 15:17 Mon 22 Nov     , Kelly O'Hair wrote:
> Been scratching my head on this one.
> 
> Another error is in the target sane-gcc-compiler, same kind of thing.
> Turns out this one was Solaris only, and the older version of GNU make  
> seemed to be silent (3.78.1)
> but the newer one (3.81) complained about a missing TAB character.
> Our RE teams use the older GNU make 3.78.1, I've been trying to  
> advance (if you want to
> call it that) to 3.81, or even 3.82 if possible, but it's not been as  
> easy as originally planned
> due to the issues with Windows, MKS, and escaped quotes in rules. But  
> 3.82 may have a fix
> for this.
> 
> Could we be using a different GNU make version on some of these Linux  
> systems?
> 

I think a move from 3.81 to 3.82 may have trigged the issue here.  I'm
don't know if Fedora 14 also upgrades to that version, but it sounds
likely.  I presume it simply wasn't been evaluated before, given there
are three bugs and two of these, the variable expansion and the
conditional, are being evaluated by the shell.  My guess would be that
it simply regarded the target as empty due to the spacing and thus it
was trivially satisfied.

> -kto
> 
> On Nov 22, 2010, at 2:33 PM, David Holmes wrote:
> 
> > Any enlightenment on how this only just started happening? Has the  
> > alsa check been disabled previously?
> >
> > David
> >
> > Kelly O'Hair said the following on 11/23/10 08:06:
> >> On Nov 22, 2010, at 1:53 PM, Dr Andrew John Hughes wrote:
> >>> On 22:15 Mon 22 Nov     , Patrick Reinhart wrote:
> >>>> Am 22.11.10 22:09, schrieb Dr Andrew John Hughes:
> >>>>> I'm quite puzzled as to how this hasn't been spotted before now,  
> >>>>> but I
> >>>>> tried to build jdk7/jdk7 today (b118 from hg), using exactly the  
> >>>>> same
> >>>>> script as I usually do, and immediately failed due to a missing
> >>>>> separator in the jdk Sanity.gmk Makefile:
> >>>>>
> >>>>> make[1]: Entering directory `/home/andrew/projects/openjdk/ 
> >>>>> upstream/build/jdk/make'
> >>>>> /home/andrew/projects/openjdk/upstream/build/jdk/make/common/ 
> >>>>> shared/Sanity.gmk:1392: *** missing separator (did you mean TAB  
> >>>>> instead of 8 spaces?).  Stop.
> >>>>> make[1]: Leaving directory `/home/andrew/projects/openjdk/ 
> >>>>> upstream/build/jdk/make'
> >>>>>
> >>>>> I've just confirmed it's also broken in the build tree and the  
> >>>>> icedtea tree
> >>>>> also fails (which is where I first hit the error).
> >>>>>
> >>>>> It seems the sane-alsa-headers target is completely broken.   
> >>>>> What's puzzling
> >>>>> is the changes occur in:
> >>>>>
> >>>>> http://hg.openjdk.java.net/jdk7/jdk7/jdk/rev/30bf00392b6d
> >>>>> changeset:   914:30bf00392b6d
> >>>>> parent:      796:d8eb2738db6b
> >>>>> user:        ohair
> >>>>> date:        Sat Jan 31 17:31:21 2009 -0800
> >>>>> summary:     6799141: Build with --hash-style=both so that  
> >>>>> binaries can work on SuSE 10
> >>>>>
> >>>>> which is nearly two years old.
> >>>>>
> >>>>> I managed to make some headway:
> >>>>>
> >>>>> * The target uses a mix of tabs and spaces.  Replacing with all  
> >>>>> tabs gets things further.
> >>>>> * The next issue is fixed by changing $${alsa_version) to $$ 
> >>>>> {alsa_version}.
> >>>>>
> >>>>> It then fails because we have a conditional structured as:
> >>>>>
> >>>>> if [ "$(ALSA_CHECK)" != "same" -a "$(ALSA_CHECK)" !=  
> >>>>> "newer"  ] ; then \
> >>>>> ...
> >>>>> fi \
> >>>>> else \
> >>>>> ...
> >>>>> fi
> >>>>>
> >>>>> There seems to be an if statement missing as, otherwise, having  
> >>>>> REQUIRED_ALSA_VERSION
> >>>>> defined always results in an error.  I assume there should be a  
> >>>>> top level if statement,
> >>>>> similar to the @if [ -f "$(ALSA_VERSION_CHECK)" ]; removed by  
> >>>>> this changeset.  This would
> >>>>> also explain why the if block is printed when usually such  
> >>>>> things are silent.
> >>>>>
> >>>>> Anyone care to enlighten us as to the missing if statement?  I'd  
> >>>>> also love to know how
> >>>>> this has only just started biting me now.
> >>>>>
> >>>>> Thanks,
> >>>> Hi Andrew,
> >>>>
> >>>> See my statement and correcting diff
> >>>> http://mail.openjdk.java.net/pipermail/build-dev/2010-November/003578.html
> >>>> when I tried to get the build running under Fedora 14...
> >>>>
> >>>> Regards Patrick
> >>>>
> >>>>
> >>>
> >>> Yes, as described in my original e-mail, I've already made those  
> >>> changes.
> >>>
> >>> I don't believe the final else block is redundant.  An if check  
> >>> seems to be missing, which
> >>> is further reinforced if you take a look at other checks in the  
> >>> file and the CheckVersions macro;
> >>> we should be looking for ALSA_CHECK being "missing".
> >>>
> >>> I've posted a webrev here: http://cr.openjdk.java.net/~andrew/build/webrev.07/
> >>>
> >>> which contains corrected indenting, the alsa_version fix and the  
> >>> additional if test.
> >>>
> >>> Kelly, does this look ok to push? If so, can I have a bug ID for it?
> >> Looks fine. Feel free to push.
> >> 7000225: Sanity check on sane-alsa-headers is broken
> >> Thank you. I filed the bug last week I think, just been so busy  
> >> lately. :^(
> >> -kto
> >>>
> >>> Thanks,
> >>> -- 
> >>> Andrew :)
> >>>
> >>> Free Java Software Engineer
> >>> Red Hat, Inc. (http://www.redhat.com)
> >>>
> >>> Support Free Java!
> >>> Contribute to GNU Classpath and the OpenJDK
> >>> http://www.gnu.org/software/classpath
> >>> http://openjdk.java.net
> >>> PGP Key: 94EFD9D8 (http://subkeys.pgp.net)
> >>> Fingerprint = F8EF F1EA 401E 2E60 15FA  7927 142C 2591 94EF D9D8
> 

-- 
Andrew :)

Free Java Software Engineer
Red Hat, Inc. (http://www.redhat.com)

Support Free Java!
Contribute to GNU Classpath and the OpenJDK
http://www.gnu.org/software/classpath
http://openjdk.java.net
PGP Key: 94EFD9D8 (http://subkeys.pgp.net)
Fingerprint = F8EF F1EA 401E 2E60 15FA  7927 142C 2591 94EF D9D8