Grant Grundler wrote: > Later in the patch I see: > + pm8001_ha->tags = kmalloc(sizeof(unsigned long)*PM8001_MAX_DEVICES, > + GFP_KERNEL); > > What are we limiting to 1024 devices? LUNS? > I noticed the constant was also used in an initialization loop too. There are some other things that I find suspicious at this point: -for allocations of array kcalloc() is usually the way to go as it does overflow checks for the multiplication result (not needed here with the current values, but doesn't really hurt in init path either) -using sizeof(*pm8001_ha->tags) would make sure it's the correct size no matter what type this will ever be, because ... -unsigned long has different size on 32 and 64 bit, I don't know if that makes sense here. AFAICT this is sort of bitmap here so this probably doesn't need to double it's size on 64 bit. Greetings, Eike
Attachment:
signature.asc
Description: This is a digitally signed message part.
