#2268 closed defect (fixed)
Ingr 1.5.0 Runtime failure on Solaris 10
Reported by: | PeterHalls | Owned by: | ilucena |
---|---|---|---|
Priority: | normal | Milestone: | 1.5.1 |
Component: | GDAL_Raster | Version: | 1.5.0 |
Severity: | normal | Keywords: | INGR Sparc Solaris |
Cc: | warmerdam, Mateusz Łoskot |
Description
When GDAL 1.5.0 is compiled on Sun Solaris (Sparc) 10 using the Sun Studio compiler version 5.9 Patch 124867-02 dated 2007.11.27. Programs linked to OGR fail at runtime with "Bus error (core dumped)". Running debugger on the core dump shows the failure to be "invalid address alignment" in SLIP.INIT_A called from IngrTypes.cpp line 64.
There are no such problems compiling with GCC.
Change History (3)
comment:1 by , 15 years ago
Cc: | added |
---|---|
Component: | default → GDAL_Raster |
Keywords: | INGR Sparc Solaris added |
Milestone: | → 1.5.2 |
Owner: | changed from | to
comment:2 by , 15 years ago
Resolution: | → fixed |
---|---|
Status: | new → closed |
comment:3 by , 15 years ago
Milestone: | 1.5.2 → 1.5.1 |
---|
I have taken the liberty of back porting this patch as r13972 in 1.5 branch in time for the 1.5.1 release.
Note:
See TracTickets
for help on using tickets.
Ivan,
I looked at line 64 of IngrTypes.cpp and it seems to be in the middle of the static initializer for a structure so I'm not sure that aspects tells us much. I don't know if there is much you can do on this without access to the system.
I will say that Sparc's are quick sensitive about pointer alignment. For instance, if you access through a double pointer that isn't on an 8byte memory boundary it will trigger an exception. I generally work around this with memcpy()s when pulling from potentially unaligned read buffers.
If you can think of a possible problem, I think Peter may be able to test a potential fix. If not, we might be able to get access to a Sparc test system at Telascience.