This is the mail archive of the
ecos-discuss@sources.redhat.com
mailing list for the eCos project.
Re: Bug in getifaddrs() - allocation of buffer
- From: Gary Thomas <gary at mlbassoc dot com>
- To: Andrew Lunn <andrew at lunn dot ch>
- Cc: Jay Foster <jay dot foster at systech dot com>,"'ecos-discuss at sources dot redhat dot com'" <ecos-discuss at sources dot redhat dot com>
- Date: 16 Sep 2003 10:55:34 -0600
- Subject: Re: [ECOS] Bug in getifaddrs() - allocation of buffer
- Organization: MLB Associates
- References: <80B97DE95AEED311BA580050047FE98494F6EC@mail.systech.com> <20030916163717.GA16899@londo.lunn.ch>
On Tue, 2003-09-16 at 10:37, Andrew Lunn wrote:
> On Tue, Sep 16, 2003 at 08:18:11AM -0700, Jay Foster wrote:
> > The function getifaddrs() in file packages/net/common/current/src/ifaddrs.c
> > has a couple of bugs in it.
> >
> > On lines 151 and 157, it calls "free(buf)", but "buf" is a local buffer
> > allocated on the stack. These should either be removed, or change the
> > allocation of "buf" to be malloced. These free() calls seem to indicate
> > that at one time, the "buf" buffer was malloced, which makes sense, since
> > it is a 1K byte buffer. This seems a bit large for an auto stack variable.
>
> I agree, it should be malloc'd. Its been broken since at least
> 20-May-02!
>
Actually, it's been broken (in this respect) forever :-( This is how
the code looked when originally imported.
> >
> > On lines 172 through 177 (CYGPKG_NET_INET6), it returns from the function
> > (error case), but does not free the malloced buffer "data". Need to add
> > a call to "free(data);" here. Also, if the "buf" buffer allocation is
> > changed to be malloced, instead of on the stack, then add a "free(buf);"
> > here too.
>
> Agreed.
Yes, a little looking at this file and how those pointers [objects] are
allocated and freed would be a good idea.
--
Gary Thomas <gary@mlbassoc.com>
MLB Associates
--
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss