Customer reports:

The attached file makes problems with GS 8.53, tiffcmyk device:                
                                                                               
- at 1519 dpi it exits with "/VMError in --fill--"                              
- at 1634 dpi it crashes with SIGSEGV                                           
                                                ther problems of this type      
The /VMError comes from gs_heap_alloc_bytes failing to alloc a large bloc       
of memory (something like 3,9 Gb). The allocation comes from                    
gx_pattern_load() / pattern_accum_open()                                        
                                                                                
The crash is caused again by an overflow ( pattern_accum_open would have to
claim 4,5 Gb of memory, but this becomes a few megabytes only in an
unsigned int variable, so the allocation is OK but                             
mem_true32_fill_rectangle crashes immediately).                                
                                                                               
If I remember correcly, I have already seem this bug reported on the Bug        
Tracker, and it's due to the inability of GS to make texture mapping           
"on-the-fly", or something similar. Nevertheless, I submit you the file in     
case it's something else (it has some spot colors, so maybe it's the           
combination spot/textures).                                                     
                                                                               
It seems this time I can't anything, but just one question: do I have a        
sure way to know (either in the device procs or from somwehere in the core     
code), when such a file is ripped at low resolution, that it is likely to       
cause this bug at high resolution? I could then, at least, warn the user        
when he attempt to print it, after the preview).