On Thu, Jan 22, 2009 at 10:48:59AM +0000, Daniel P. Berrange wrote:
On Fri, Jan 16, 2009 at 12:25:02PM +0000, Daniel P. Berrange wrote:
> On Fri, Jan 16, 2009 at 12:21:59PM +0000, Richard W.M. Jones wrote:
> > On Tue, Jan 13, 2009 at 05:46:37PM +0000, Daniel P. Berrange wrote:
> > > + char buf[1024];
> >
> > sysconf (_SC_GETPW_R_SIZE_MAX)?
> >
> > Looking at glibc's implementation of getpwuid (which uses getpwuid_r),
> > I see that glibc dynamically reallocates the buffer as necessary to
> > the correct size for the return value. The logic of this is fairly
> > simple so maybe we should do the same?
> >
> > From glibc:
> >
> > buffer = malloc (/*some_initial_size*/);
> >
> > while (buffer != NULL
> > && (getpwuid_r (const char *name, &resbuf, buffer,
> > buffer_size, &result)
> > == ERANGE))
> > {
> > char *new_buf;
> > buffer_size *= 2;
> > new_buf = (char *) realloc (buffer, buffer_size);
> > if (new_buf == NULL)
> > {
> > free (buffer);
> > errno = ENOMEM;
> > }
> > buffer = new_buf;
> > }
> >
> >
> > Anyhow, +1 but I'd be happier if these functions were centralized in
> > somewhere like src/utils.c.
>
> That's a good idea - in all the cases where we currently use getpwuid
> all we actually want is the home directory path. So we could add a
> simple func:
>
> char *virUserHomeDirectory(uid_t uid);
>
> and hide all the horrific code in there.
Here's an update with that usage in it
That's better. There's still the question about whether the
particular sysconf() sub-call we are using exists on all platforms,
and we might need to replace it with the glibc-style incremental
buffer.
I say we should commit this and test it for a while.
Rich.
--
Richard Jones, Emerging Technologies, Red Hat
http://et.redhat.com/~rjones
virt-p2v converts physical machines to virtual machines. Boot with a
live CD or over the network (PXE) and turn machines into Xen guests.
http://et.redhat.com/~rjones/virt-p2v