Re: memory leak in generic_parse_monolithic [+PATCH]

[David Howells]

Randy Dunlap <rdunlap@xxxxxxxxxxxxx> wrote:

> Here's the syzbot reproducer:
> https://syzkaller.appspot.com/x/repro.c?x=129ca3d6500000
> 
> The "interesting" mount params are:
> 	source=%^]$[+%](${:\017k[)-:,source=%^]$[+.](%{:\017\200[)-:,\000
> 
> There is no other AFS activity: nothing mounted, no cells known (or
> whatever that is), etc.
> 
> I don't recall if the mount was successful and I can't test it just now.
> My laptop is mucked up.
> 
> 
> Be aware that this report could just be a false positive: it waits
> for 5 seconds then looks for a memleak. AFAIK, it's possible that the "leaked"
> memory is still in valid use and will be freed some day.

Bah.  Multiple source= parameters.  I don't reject the second one, but just
overwrite fc->source.

David