linux-nfs.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Bruce Fields <bfields@fieldses.org>
To: Simo Sorce <simo@redhat.com>
Cc: Chuck Lever <chuck.lever@oracle.com>,
	Linux NFS Mailing List <linux-nfs@vger.kernel.org>
Subject: Re: [PATCH v1 2/2] SUNRPC: Fix svcauth_gss_proxy_init()
Date: Wed, 30 Oct 2019 16:33:37 -0400	[thread overview]
Message-ID: <20191030203337.GA13537@fieldses.org> (raw)
In-Reply-To: <59d0188a2195f7f8b2416179a7cc8813a147939f.camel@redhat.com>

On Thu, Oct 24, 2019 at 10:02:27AM -0400, Simo Sorce wrote:
> On Thu, 2019-10-24 at 09:35 -0400, Chuck Lever wrote:
> > Whoops, was going to Cc: Simo on this one...
> 
> Doesn't look like it is changing any behavior wrt GSS, so fine with me.

Thanks, Simo.  I've added a Reviewed-by: Simo Sorce <simo@redhat.com>
to this one.

--b.

> 
> Simo.
> 
> > > On Oct 24, 2019, at 9:34 AM, Chuck Lever <chuck.lever@oracle.com> wrote:
> > > 
> > > gss_read_proxy_verf() assumes things about the XDR buffer containing
> > > the RPC Call that are not true for buffers generated by
> > > svc_rdma_recv().
> > > 
> > > RDMA's buffers look more like what the upper layer generates for
> > > sending: head is a kmalloc'd buffer; it does not point to a page
> > > whose contents are contiguous with the first page in the buffers'
> > > page array. The result is that ACCEPT_SEC_CONTEXT via RPC/RDMA has
> > > stopped working on Linux NFS servers that use gssproxy.
> > > 
> > > This does not affect clients that use only TCP to send their
> > > ACCEPT_SEC_CONTEXT operation (that's all Linux clients). Other
> > > clients, like Solaris NFS clients, send ACCEPT_SEC_CONTEXT on the
> > > same transport as they send all other NFS operations. Such clients
> > > can send ACCEPT_SEC_CONTEXT via RPC/RDMA.
> > > 
> > > I thought I had found every direct reference in the server RPC code
> > > to the rqstp->rq_pages field.
> > > 
> > > Bug found at the 2019 Westford NFS bake-a-thon.
> > > 
> > > Fixes: 3316f0631139 ("svcrdma: Persistently allocate and DMA- ... ")
> > > Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
> > > Tested-by: Bill Baker <bill.baker@oracle.com>
> > > ---
> > > net/sunrpc/auth_gss/svcauth_gss.c |   84 ++++++++++++++++++++++++++++---------
> > > 1 file changed, 63 insertions(+), 21 deletions(-)
> > > 
> > > diff --git a/net/sunrpc/auth_gss/svcauth_gss.c b/net/sunrpc/auth_gss/svcauth_gss.c
> > > index f130990..c62d1f1 100644
> > > --- a/net/sunrpc/auth_gss/svcauth_gss.c
> > > +++ b/net/sunrpc/auth_gss/svcauth_gss.c
> > > @@ -1078,24 +1078,32 @@ struct gss_svc_data {
> > > 	return 0;
> > > }
> > > 
> > > -/* Ok this is really heavily depending on a set of semantics in
> > > - * how rqstp is set up by svc_recv and pages laid down by the
> > > - * server when reading a request. We are basically guaranteed that
> > > - * the token lays all down linearly across a set of pages, starting
> > > - * at iov_base in rq_arg.head[0] which happens to be the first of a
> > > - * set of pages stored in rq_pages[].
> > > - * rq_arg.head[0].iov_base will provide us the page_base to pass
> > > - * to the upcall.
> > > - */
> > > -static inline int
> > > -gss_read_proxy_verf(struct svc_rqst *rqstp,
> > > -		    struct rpc_gss_wire_cred *gc, __be32 *authp,
> > > -		    struct xdr_netobj *in_handle,
> > > -		    struct gssp_in_token *in_token)
> > > +static void gss_free_in_token_pages(struct gssp_in_token *in_token)
> > > {
> > > -	struct kvec *argv = &rqstp->rq_arg.head[0];
> > > 	u32 inlen;
> > > -	int res;
> > > +	int i;
> > > +
> > > +	i = 0;
> > > +	inlen = in_token->page_len;
> > > +	while (inlen) {
> > > +		if (in_token->pages[i])
> > > +			put_page(in_token->pages[i]);
> > > +		inlen -= inlen > PAGE_SIZE ? PAGE_SIZE : inlen;
> > > +	}
> > > +
> > > +	kfree(in_token->pages);
> > > +	in_token->pages = NULL;
> > > +}
> > > +
> > > +static int gss_read_proxy_verf(struct svc_rqst *rqstp,
> > > +			       struct rpc_gss_wire_cred *gc, __be32 *authp,
> > > +			       struct xdr_netobj *in_handle,
> > > +			       struct gssp_in_token *in_token)
> > > +{
> > > +	struct kvec *argv = &rqstp->rq_arg.head[0];
> > > +	unsigned int page_base, length;
> > > +	int pages, i, res;
> > > +	size_t inlen;
> > > 
> > > 	res = gss_read_common_verf(gc, argv, authp, in_handle);
> > > 	if (res)
> > > @@ -1105,10 +1113,36 @@ struct gss_svc_data {
> > > 	if (inlen > (argv->iov_len + rqstp->rq_arg.page_len))
> > > 		return SVC_DENIED;
> > > 
> > > -	in_token->pages = rqstp->rq_pages;
> > > -	in_token->page_base = (ulong)argv->iov_base & ~PAGE_MASK;
> > > +	pages = DIV_ROUND_UP(inlen, PAGE_SIZE);
> > > +	in_token->pages = kcalloc(pages, sizeof(struct page *), GFP_KERNEL);
> > > +	if (!in_token->pages)
> > > +		return SVC_DENIED;
> > > +	in_token->page_base = 0;
> > > 	in_token->page_len = inlen;
> > > +	for (i = 0; i < pages; i++) {
> > > +		in_token->pages[i] = alloc_page(GFP_KERNEL);
> > > +		if (!in_token->pages[i]) {
> > > +			gss_free_in_token_pages(in_token);
> > > +			return SVC_DENIED;
> > > +		}
> > > +	}
> > > 
> > > +	length = min_t(unsigned int, inlen, argv->iov_len);
> > > +	memcpy(page_address(in_token->pages[0]), argv->iov_base, length);
> > > +	inlen -= length;
> > > +
> > > +	i = 1;
> > > +	page_base = rqstp->rq_arg.page_base;
> > > +	while (inlen) {
> > > +		length = min_t(unsigned int, inlen, PAGE_SIZE);
> > > +		memcpy(page_address(in_token->pages[i]),
> > > +		       page_address(rqstp->rq_arg.pages[i]) + page_base,
> > > +		       length);
> > > +
> > > +		inlen -= length;
> > > +		page_base = 0;
> > > +		i++;
> > > +	}
> > > 	return 0;
> > > }
> > > 
> > > @@ -1282,8 +1316,11 @@ static int svcauth_gss_proxy_init(struct svc_rqst *rqstp,
> > > 		break;
> > > 	case GSS_S_COMPLETE:
> > > 		status = gss_proxy_save_rsc(sn->rsc_cache, &ud, &handle);
> > > -		if (status)
> > > +		if (status) {
> > > +			pr_info("%s: gss_proxy_save_rsc failed (%d)\n",
> > > +				__func__, status);
> > > 			goto out;
> > > +		}
> > > 		cli_handle.data = (u8 *)&handle;
> > > 		cli_handle.len = sizeof(handle);
> > > 		break;
> > > @@ -1294,15 +1331,20 @@ static int svcauth_gss_proxy_init(struct svc_rqst *rqstp,
> > > 
> > > 	/* Got an answer to the upcall; use it: */
> > > 	if (gss_write_init_verf(sn->rsc_cache, rqstp,
> > > -				&cli_handle, &ud.major_status))
> > > +				&cli_handle, &ud.major_status)) {
> > > +		pr_info("%s: gss_write_init_verf failed\n", __func__);
> > > 		goto out;
> > > +	}
> > > 	if (gss_write_resv(resv, PAGE_SIZE,
> > > 			   &cli_handle, &ud.out_token,
> > > -			   ud.major_status, ud.minor_status))
> > > +			   ud.major_status, ud.minor_status)) {
> > > +		pr_info("%s: gss_write_resv failed\n", __func__);
> > > 		goto out;
> > > +	}
> > > 
> > > 	ret = SVC_COMPLETE;
> > > out:
> > > +	gss_free_in_token_pages(&ud.in_token);
> > > 	gssp_free_upcall_data(&ud);
> > > 	return ret;
> > > }
> > > 
> > 
> > --
> > Chuck Lever
> > 
> > 
> > 
> 
> -- 
> Simo Sorce
> RHEL Crypto Team
> Red Hat, Inc
> 
> 
> 

  reply	other threads:[~2019-10-30 20:33 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-10-24 13:34 [PATCH v1 1/2] SUNRPC: Trace gssproxy upcall results Chuck Lever
2019-10-24 13:34 ` [PATCH v1 2/2] SUNRPC: Fix svcauth_gss_proxy_init() Chuck Lever
2019-10-24 13:35   ` Chuck Lever
2019-10-24 14:02     ` Simo Sorce
2019-10-30 20:33       ` Bruce Fields [this message]
2019-10-24 15:38 ` [PATCH v1 1/2] SUNRPC: Trace gssproxy upcall results J. Bruce Fields
2019-10-24 17:08   ` Chuck Lever
2019-10-28 16:45     ` Bruce Fields
2019-10-30 20:34       ` Bruce Fields

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20191030203337.GA13537@fieldses.org \
    --to=bfields@fieldses.org \
    --cc=chuck.lever@oracle.com \
    --cc=linux-nfs@vger.kernel.org \
    --cc=simo@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).