Bug 100050 - lcms 1.18 update breaks ICC_ProfileRGB Tests

[Andrew John Hughes]

2009/5/8 Andrew Haley <aph at redhat.com>:
> Colour management is broken at the moment because the import of the
> latest OpenJDK 6 contains a new version of lcms that is buggy.
>
> I've written a patch to fix this, and Sun are looking at it.  In the
> meantime I propose to patch IcedTea as a temporary workaround.
>
> OK?
>
> Andrew.
>
>
>
> --- LCMS.c~     2009-04-24 08:34:31.000000000 +0100
> +++ LCMS.c      2009-05-07 19:11:37.000000000 +0100
> @@ -610,14 +610,19 @@
>         return FALSE;
>     }
>
> -    if (!Icc->Grow(Icc, delta)) {
> -        free(ptr);
> -        if(isNew) {
> -            Icc->TagCount--;
> -        }
> -        J2dRlsTraceLn(J2D_TRACE_ERROR,
> -                      "_cmsModifyTagData: Icc->Grow() == FALSE");
> -        return FALSE;
> +    /* We change the size of Icc here only if we know it'll actually
> +     * grow: if Icc is about to shrink we must wait until we've read
> +     * the previous data.  */
> +    if (delta > 0) {
> +       if (!Icc->Grow(Icc, delta)) {
> +           free(ptr);
> +           if(isNew) {
> +               Icc->TagCount--;
> +           }
> +           J2dRlsTraceLn(J2D_TRACE_ERROR,
> +                         "_cmsModifyTagData: Icc->Grow() == FALSE");
> +           return FALSE;
> +       }
>     }
>
>     /* Compute size of tag data before/after the modified tag */
> @@ -680,6 +685,18 @@
>     temp = TransportValue32(profileSize);
>     Icc->Write(Icc, sizeof(icUInt32Number), &temp);
>
> +    /* Shrink Icc, if needed.  */
> +    if (delta < 0) {
> +       if (!Icc->Grow(Icc, delta)) {
> +           free(ptr);
> +           if(isNew) {
> +               Icc->TagCount--;
> +           }
> +           J2dRlsTraceLn(J2D_TRACE_ERROR,
> +                         "_cmsModifyTagData: Icc->Grow() == FALSE");
> +           return FALSE;
> +       }
> +    }
>
>     /* Adjust tag offsets: if the tag is new, we must account
>        for the new tag table entry; otherwise, only those tags after
>

Sounds sensible to me.  Can you make sure to add it to the HACKING
file/wiki page with any tracking data like bug IDs so we know that
this is already on its way upstream?

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