Re: First release candidate for GDB 6.4 available

[Wu Zhou <woodzltc at cn dot ibm dot com>]

Hi Mark,

On Thu, 24 Nov 2005, Mark Kettenis wrote:

> > Date: Wed, 23 Nov 2005 11:54:40 +0800 (CST)
> > From: Wu Zhou <woodzltc@cn.ibm.com>
> > 
> > I run this RC on ppc64 platform as both 32-bits and 64-bits application.
> > Here is the summary.  I also list the x86 summary as reference.
> > 
> > ppc64, 64-bits gdb, 32-bits testcase:
> > ======================================
> ...
> > FAIL: gdb.base/recurse.exp: continue to recurse (a = 8)
> > FAIL: gdb.base/recurse.exp: continue to recurse (a = 7)
> > FAIL: gdb.base/recurse.exp: continue to recurse (a = 6)
> > FAIL: gdb.base/recurse.exp: continue to recurse (a = 5)
> > FAIL: gdb.base/recurse.exp: continue to second instance watchpoint, first time
> > FAIL: gdb.base/recurse.exp: continue to recurse (a = 4)
> > FAIL: gdb.base/recurse.exp: continue to recurse (a = 3)
> > FAIL: gdb.base/recurse.exp: continue to recurse (a = 2)
> > FAIL: gdb.base/recurse.exp: continue to recurse (a = 1)
> > FAIL: gdb.base/recurse.exp: continue to second instance watchpoint, second time

The mainline GDB-6.3 only report three failures here. And RHEL4's GDB, 
which is based on 6.3.0 and added some of their own patch,  report all ok. 
So I believe that these are regression. 

BTW. I am also tracking these problems and guessing that it might be 
related to the software watchpoint code. I did find the code to single 
step the inferior, but I didn't find the code to check whether these 
watched variables get changed or not. (These code seems somewaht messy to 
me, :-) Would you please give some pointer?  Thanks in advance!

> ...
> > FAIL: gdb.cp/anon-union.exp: print w 1
> > FAIL: gdb.cp/anon-union.exp: print z 1
> > FAIL: gdb.cp/anon-union.exp: print w 2
> > FAIL: gdb.cp/anon-union.exp: print z 2
> > FAIL: gdb.cp/anon-union.exp: print w 3
> > FAIL: gdb.cp/anon-union.exp: print z 3

I guess that GDB don't know how to handle anonymous union yet.  I ever saw 
these failures in a few platform.  And also see them ok on others.  The 
difference I find is that some version of gcc handle the anonymous union 
members as normal variables, others don't. 

> > FAIL: gdb.gdb/observer.exp: second observer attached; check second observer counter value
> > FAIL: gdb.gdb/observer.exp: 1st observer added; check first observer counter value
> > FAIL: gdb.gdb/observer.exp: 2nd observer added; check first observer counter value
> > FAIL: gdb.gdb/observer.exp: 2nd observer added; check second observer counter value
> > FAIL: gdb.gdb/observer.exp: 3rd observer added; check first observer counter value
> > FAIL: gdb.gdb/observer.exp: 3rd observer added; check second observer counter value
> > FAIL: gdb.gdb/observer.exp: 3rd observer added; check third observer counter value
> > FAIL: gdb.gdb/observer.exp: 2nd observer removed; check first observer counter value
> > FAIL: gdb.gdb/observer.exp: 2nd observer removed; check third observer counter value
> > FAIL: gdb.gdb/observer.exp: 1st observer removed; check third observer counter value
> > FAIL: gdb.gdb/observer.exp: three observers added; check first observer counter value
> > FAIL: gdb.gdb/observer.exp: three observers added; check second observer counter value
> > FAIL: gdb.gdb/observer.exp: three observers added; check third observer counter value
> > FAIL: gdb.gdb/observer.exp: third observer removed; check first observer counter value
> > FAIL: gdb.gdb/observer.exp: third observer removed; check second observer counter value
> > FAIL: gdb.gdb/observer.exp: second observer removed; check first observer counter value
> ...

I remember saw these error with an earlier gdb version this year.  But 
don't take any looks into them.  GDB-6.3 testsuite report "Couldn't test 
self" and then exit on my POWER4 box. Don't know why.  Will take some look 
into them later.

> For OpenBSD, the failures above are regressions from 6.3, and I think
> it's safe to assume they're regressions for Linux too.  They were
> still OK in 6.3.50.20050911-cvs, and I think remembering I've already
> seen them fail in early november.  Don't know if this is serious
> enough to delay the release.  I'll try to track this down, and
> hopefully will find the problem.  If it's too late for 6.4 we should
> release 6.4.1 with the fix.

That is quite reasonable to me.

BTW. If you have any patch, please let me know.  I am very happy to try 
them.

Regards
- Wu Zhou