Reviewer needed - fix for PR1018: JVM fails due to SEGV during rendering some Unicode characters

[Andrew Hughes]

----- Original Message -----
> ----- Original Message -----
> > Andrew Hughes wrote:
> > > 
> > > ----- Original Message -----
> > >> ----- Original Message -----
> > >>> Andrew Hughes wrote:
> > >>>> ----- Original Message -----
> > >>>>> ----- Original Message -----
> > >>>>>> Hi,
> > >>>>>>
> > >>>>>> I've created a fix for PR1018: JVM fails due to SEGV during
> > >>>>>> rendering
> > >>>>>> some Unicode characters.
> > >>>>>> This fix is based on OpenJDK7 sources and it's been
> > >>>>>> successfully
> > >>>>>> tested against IcedTea6-head,
> > >>>>>> IcedTea6-1.11 and IcedTea6-1.10.
> > >>>>>>
> > >>>>> You say this is based on "OpenJDK7 sources".  What fix is
> > >>>>> this
> > >>>>> from
> > >>>>> and
> > >>>>> why is there no bug reference in this patch?
> > >>>> This appears to be a fragment of:
> > >>> yes, it is
> > >>>
> > >>>> changeset:   3068:1d4340015b85
> > >>>> user:        srl
> > >>>> date:        Mon Dec 06 16:10:01 2010 -0800
> > >>>> summary:     6886358: layout code update
> > >>>>
> > >>>> Why not apply the whole fix?  Including chunks like this makes
> > >>>> things
> > >>>> untraceable and harder to upstream.
> > >>> well this fix is 880 kilobytes long and it changes a lot of
> > >>> code.
> > >>> I'm
> > >>> not fully
> > >>> against to use the whole fix, but I'm a bit scared if it does
> > >>> introduce
> > >>> some new bugs into a now-stable IcedTea6.
> > >> A large part of that is regenerating machine-generated
> > >> substitution
> > >> tables.
> > >> There are also two new tests in there, and this code has been in
> > >> 7
> > >> for nearly
> > >> eighteen months.  If there are issues, they may also have been
> > >> fixed
> > >> in 7.
> > >>
> > >> On the contrary, this subsection applied on its own has not been
> > >> tested in the
> > >> wild at all.
> > >>
> > >> Is this urgent for 6?  Maybe we could compromise and put the
> > >> full
> > >> fix
> > >> in HEAD,
> > >> while putting your smaller patch in 1.10 & 1.11?
> > >>
> > > 
> > > Note that for the full patch, you also need to backport:
> > > 
> > > http://hg.openjdk.java.net/jdk8/2d/jdk/rev/c892ca15ca52
> > > 
> > > as this introduces the line that causes the strict aliasing
> > > issue.
> > > 
> > Ah it's important, TY. Was this change backported to 7 too?
> > 
> 
> It went into IcedTea7 first.  Then it's taken a week to get it into 8
> (check the timestamp...).  Next target is 7u.

Oh, FYI it's in the 2.2 branch too.  Jiri approved it.

> 
> > >>>>>> Can anybody please review this fix?
> > >>>>>>
> > >>>>>> Here's ChangeLog entry:
> > >>>>>>
> > >>>>>> 2012-05-30  Pavel Tisnovsky  <ptisnovs at redhat.com>
> > >>>>>>
> > >>>>>>         * patches/coverage-table.patch:
> > >>>>>>         PR1018: JVM fails due to SEGV during rendering some
> > >>>>>>         Unicode
> > >>>>>>         characters
> > >>>>>>         * Makefile.am: Updated.
> > >>>>>>         * NEWS: Mention bugfix.
> > >>>>>>
> > >>>>>> (if this fix will be approved, I would like to backport it
> > >>>>>> to
> > >>>>>> 6-1.10
> > >>>>>> and 6-1.11 too - the
> > >>>>>> patch is the same for all branches)
> > >>>>>>
> > >>>>>> Thank you in advance,
> > >>>>>> Pavel
> > >>>>>>
> > >>>>>>
> > >>>>>> [Text Documents:coverage-table.patch]

-- 
Andrew :)

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

PGP Key: 248BDC07 (https://keys.indymedia.org/)
Fingerprint = EC5A 1F5E C0AD 1D15 8F1F  8F91 3B96 A578 248B DC07