Opened 16 years ago
Closed 16 years ago
#2535 closed defect (fixed)
JPEGDataset crash related to jpeg_start_decompress
| Reported by: | list67 | Owned by: | warmerdam |
|---|---|---|---|
| Priority: | normal | Milestone: | 1.5.3 |
| Component: | GDAL_Raster | Version: | 1.5.2 |
| Severity: | normal | Keywords: | JPEGDataset, jpeg_start_decompress |
| Cc: |
Description
In JPGDataset::Restart(), there is a jpeg_start_decompress(...) call, but there is not currently a "bHasDoneJpegStartDecompress = TRUE;" assignment there.
Adding a bHasDoneJpegStartDecompress = TRUE; assignment after that jpeg_start_decompress(...) call prevents a crash from occurring.
The crash did not occurr with gdal 1.4.2, but does occur with gdal 1.5.2.
Change History (2)
comment:1 by , 16 years ago
| Component: | default → GDAL_Raster |
|---|---|
| Milestone: | → 1.5.3 |
| Status: | new → assigned |
| Version: | svn-trunk → 1.5.2 |
comment:2 by , 16 years ago
| Resolution: | → fixed |
|---|---|
| Status: | assigned → closed |
Note:
See TracTickets
for help on using tickets.
Mike,
Change applied in trunk (r15170) and 1.5 branch (r15171). I wasn't able to reproduce the problem (I didn't really try hard. It seems it is only an issue if the RasterIO() special case is used before any block oriented io. Furthermore the rasterio special case seems to only do something special for pixeloffset > 3 which seems odd to me.