#5794 closed defect (fixed)
gdaladdo crash with certain datasets
| Reported by: | georgkwatson | Owned by: | warmerdam |
|---|---|---|---|
| Priority: | normal | Milestone: | 2.0.0 |
| Component: | Utilities | Version: | 1.11.1 |
| Severity: | major | Keywords: | |
| Cc: |
Description
If one invokes gdaladdo with a geotif of insufficient size (in one dimension or other) to generate the requested number of overviews, it crashes, and in some environments, hangs.
Let's say you have a tif of the dimensions 6x8192 (thin sliver).
Invoking: gdaladdo -r average sliver.tif 2 4 8 16 32
Will crash. This only happens with -r average, nearest Requesting cubic works. I have a fix for this that has been deployed in our production environment.
Attachments (1)
Change History (6)
by , 9 years ago
| Attachment: | sliver.tif added |
|---|
comment:1 by , 9 years ago
You didn't attached your fix, but I've commited an attempt to fix that in trunk r28292 that should work hopefully. I let it settle a bit before trying to backport in 1.11
comment:2 by , 9 years ago
My fix involved eliminating nonsensical overviews. I like yours a bit better at this point.
comment:3 by , 9 years ago
I don't feel like backporting to 1.11: too much invasive. If you have a less invasive, more local patch that just avoids the crash, I could consider it for 1.11
comment:4 by , 9 years ago
| Milestone: | 1.11.2 → 2.0 |
|---|---|
| Resolution: | → fixed |
| Status: | new → closed |
Retargetting to 2.0 as no patch has been proposed for 1.11
A skinny tif that reproduces the issue