On Thu, 2005-04-14 at 15:25 +0200, Ingo Molnar wrote:
> * Ingo Molnar <[EMAIL PROTECTED]> wrote:
> > this patch fixes a memory leak in read-cache.c: when there's cache 
> > entry collision we should free the previous one.
> > +           free(active_cache[pos]);
> >             active_cache[pos] = ce;
> i'm having second thoughs about this one: active_cache entries are not 
> always malloc()-ed - e.g. read_cache() will construct them from the 
> mmap() of the index file. Which must not be free()d!
> one safe solution would be to malloc() all these entries and copy them 
> over from the index file? Slightly slower but safer and free()-able when 
> update-cache.c notices a collision. The (tested) patch below does this.
> this would also make Martin Schlemmer's update-cache.c fix safe.

Ok, bad me it seems - assumed it was malloc'd.

Martin Schlemmer

Attachment: signature.asc
Description: This is a digitally signed message part

Reply via email to