Build pyramid crashes for this TIFF attached

[yanchen]

Inside HFADataset::IBuildOverviews function, ::GetRasterBand() returns a NULL pointer for 2th band. Number of band is not correct.

When I debugged into HFAGetRasterInfo(), I found when .aux is read, hHFA->nBands is equal to 1.

[yanchen]

data

[warmerdam]

It would be helpful to know what function, and arguments were used to create the overviews. I'm guessing some override config options are used to force use of aux/rrd overviews instead of TIFF overviews. Is that right?

[warmerdam]

In experimenting with this file, a problem was identified with EXIF IFD handling in libtiff 4. A distinct bug in the libtiff bugzilla has been created for this issue.

​http://bugzilla.maptools.org/show_bug.cgi?id=1917

I do not believe this is related to the problem you have encountered though, since I gather you are forcing HFA/AUX/RRD overview building.

[warmerdam]

Problem reproduced using the command:

 gdaladdo --config USE_RRD YES DigitizingFeatures.tif 2 4

[warmerdam]

The problem has been corrected in trunk (r14902) and 1.5 branch (r14903) by not accepting .aux files that have a different number of pixels, lines or bands than the dependent dataset in the findauxfiles function.

[warmerdam]

Added a test in trunk (r14904).