miscellaneous gss-proxy & krb5 fixes for 3.13
Subsystems (3)
These are some (what look like mostly minor) fixes prompted by some static checker results from Andi Kleen. I'm planning to queue them up for 3.13.
--b.
PATCH 1/4 svcrpc: fix gss-proxy NULL dereference in some error cases
From: "J. Bruce Fields" <[email protected]>
We depend on the xdr decoder to set this pointer, but if we error out before we decode this piece it could be left NULL.
I think this is probably tough to hit without a buggy gss-proxy.
--- net/sunrpc/auth_gss/gss_rpc_upcall.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
net/sunrpc/auth_gss/gss_rpc_upcall.c +2 -1
@@ -298,7 +298,8 @@ int gssp_accept_sec_context_upcall(struct net *net,if (res.context_handle) { data->out_handle = rctxh.exported_context_token; data->mech_oid.len = rctxh.mech.len; -memcpy (data->mech_oid.data, rctxh.mech.data, +if (rctxh.mech.data) +memcpy (data->mech_oid.data, rctxh.mech.data, data->mech_oid.len); client_name = rctxh.src_name.display_name; }
Signature · 1 line
1.7.9.5
Re: [PATCH 1/4] svcrpc: fix gss-proxy NULL dereference in some error cases
Quoted 19 lines · J. Bruce Fields
From: "J. Bruce Fields" <[email protected]>
We depend on the xdr decoder to set this pointer, but if we error out before we decode this piece it could be left NULL.
I think this is probably tough to hit without a buggy gss-proxy.
Reported-by Andi Kleen <[email protected]>Cc Simo Sorce <[email protected]>Signed-off-by J. Bruce Fields <[email protected]>--- net/sunrpc/auth_gss/gss_rpc_upcall.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)net/sunrpc/auth_gss/gss_rpc_upcall.c +0
-0
net/sunrpc/auth_gss/gss_rpc_upcall.c int gssp_accept_sec_context_upcall(struct net *net, · line 305
Quoted 9 lines
net/sunrpc/auth_gss/gss_rpc_upcall.c +2
-1@@ -298,7 +298,8 @@ int gssp_accept_sec_context_upcall(struct net *net,if (res.context_handle) { data->out_handle = rctxh.exported_context_token; data->mech_oid.len = rctxh.mech.len; -memcpy (data->mech_oid.data, rctxh.mech.data, +if (rctxh.mech.data) +memcpy (data->mech_oid.data, rctxh.mech.data, data->mech_oid.len); client_name = rctxh.src_name.display_name; }
Signature · 1 line
Simo Sorce * Red Hat, Inc * New York
PATCH 2/4 svcrpc: fix error-handling on badd gssproxy downcall
In reply to J. Bruce FieldsFrom: "J. Bruce Fields" <[email protected]>
For every other problem here we bail out with an error, but here for some reason we're setting a negative cache entry (with, note, an undefined expiry).
It seems simplest just to bail out in the same way as we do in other cases.
--- net/sunrpc/auth_gss/svcauth_gss.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
net/sunrpc/auth_gss/svcauth_gss.c +2 -2
@@ -1167,8 +1167,8 @@ static int gss_proxy_save_rsc(struct cache_detail *cd,if (!ud->found_creds) {/* userspace seem buggy, we should always get at least a * mapping to nobody */ -dprintk ("RPC: No creds found, marking Negative!\n" ); -set_bit (CACHE_NEGATIVE , &rsci.h.flags); +dprintk ("RPC: No creds found!\n" ); + goto out; }else {/* steal creds */
Signature · 1 line
1.7.9.5
Re: [PATCH 2/4] svcrpc: fix error-handling on badd gssproxy downcall
Quoted 21 lines · J. Bruce Fields
From: "J. Bruce Fields" <[email protected]>
For every other problem here we bail out with an error, but here for some reason we're setting a negative cache entry (with, note, an undefined expiry).
It seems simplest just to bail out in the same way as we do in other cases.
Cc Simo Sorce <[email protected]>Reported-by Andi Kleen <[email protected]>Signed-off-by J. Bruce Fields <[email protected]>--- net/sunrpc/auth_gss/svcauth_gss.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)net/sunrpc/auth_gss/svcauth_gss.c +0
-0
net/sunrpc/auth_gss/svcauth_gss.c static int gss_proxy_save_rsc(struct cache_detail *cd, · line 1174
Quoted 10 lines
net/sunrpc/auth_gss/svcauth_gss.c +2
-2@@ -1167,8 +1167,8 @@ static int gss_proxy_save_rsc(struct cache_detail *cd,if (!ud->found_creds) {/* userspace seem buggy, we should always get at least a * mapping to nobody */ -dprintk ("RPC: No creds found, marking Negative!\n" ); -set_bit (CACHE_NEGATIVE , &rsci.h.flags); +dprintk ("RPC: No creds found!\n" ); + goto out; }else {/* steal creds */
IIRC, we are doing this to avoid rapid upcall loops in the kernel, where we keep hammering upcalls out and keep getting an error back. I am not sure it is wise to not set a temp negative cache here.
Simo.
Signature · 1 line
Simo Sorce * Red Hat, Inc * New York
Re: [PATCH 2/4] svcrpc: fix error-handling on badd gssproxy downcall
Quoted 31 lines · Simo Sorce
Quoted 30 lines · J. Bruce Fields
From: "J. Bruce Fields" <[email protected]>
For every other problem here we bail out with an error, but here for some reason we're setting a negative cache entry (with, note, an undefined expiry).
It seems simplest just to bail out in the same way as we do in other cases.
Cc Simo Sorce <[email protected]>Reported-by Andi Kleen <[email protected]>Signed-off-by J. Bruce Fields <[email protected]>--- net/sunrpc/auth_gss/svcauth_gss.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)net/sunrpc/auth_gss/svcauth_gss.c +2
-2@@ -1167,8 +1167,8 @@ static int gss_proxy_save_rsc(struct cache_detail *cd,if (!ud->found_creds) {/* userspace seem buggy, we should always get at least a * mapping to nobody */ -dprintk ("RPC: No creds found, marking Negative!\n" ); -set_bit (CACHE_NEGATIVE , &rsci.h.flags); +dprintk ("RPC: No creds found!\n" ); + goto out; }else {
Quoted 4 lines
net/sunrpc/auth_gss/svcauth_gss.c static int gss_proxy_save_rsc(struct cache_detail *cd, · line 1174
Quoted 1 line
net/sunrpc/auth_gss/svcauth_gss.c +0
-0/* steal creds */ IIRC, we are doing this to avoid rapid upcall loops in the kernel, where we keep hammering upcalls out and keep getting an error back.
Looks like returning an error instead results in closing the connection to the client, so, depends how the client replies I guess.
In any case I don't see why we'd treat this particular gss-proxy bug differently than we would any other (like, say, passing down bad xdr, or a gss context that we can't import).
--b.
Quoted 1 line
I am not sure it is wise to not set a temp negative cache here.
Re: [PATCH 2/4] svcrpc: fix error-handling on badd gssproxy downcall
Quoted 37 lines · J. Bruce Fields
Quoted 31 lines · Simo Sorce
Quoted 30 lines · J. Bruce Fields
From: "J. Bruce Fields" <[email protected]>
For every other problem here we bail out with an error, but here for some reason we're setting a negative cache entry (with, note, an undefined expiry).
It seems simplest just to bail out in the same way as we do in other cases.
Cc Simo Sorce <[email protected]>Reported-by Andi Kleen <[email protected]>Signed-off-by J. Bruce Fields <[email protected]>--- net/sunrpc/auth_gss/svcauth_gss.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)net/sunrpc/auth_gss/svcauth_gss.c +2
-2@@ -1167,8 +1167,8 @@ static int gss_proxy_save_rsc(struct cache_detail *cd,if (!ud->found_creds) {/* userspace seem buggy, we should always get at least a * mapping to nobody */ -dprintk ("RPC: No creds found, marking Negative!\n" ); -set_bit (CACHE_NEGATIVE , &rsci.h.flags); +dprintk ("RPC: No creds found!\n" ); + goto out; }else {Quoted 4 lines
net/sunrpc/auth_gss/svcauth_gss.c static int gss_proxy_save_rsc(struct cache_detail *cd, · line 1174
Quoted 1 line
net/sunrpc/auth_gss/svcauth_gss.c +0
-0/* steal creds */ IIRC, we are doing this to avoid rapid upcall loops in the kernel, where we keep hammering upcalls out and keep getting an error back.
Quoted 6 lines
Looks like returning an error instead results in closing the connection to the client, so, depends how the client replies I guess.
In any case I don't see why we'd treat this particular gss-proxy bug differently than we would any other (like, say, passing down bad xdr, or a gss context that we can't import).
Uhmm I did not recall that, I guess the change is ok then.
Simo.
Quoted 7 lines
--b.
Quoted 1 line
I am not sure it is wise to not set a temp negative cache here.
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Signature · 1 line
Simo Sorce * Red Hat, Inc * New York
Re: [PATCH 2/4] svcrpc: fix error-handling on badd gssproxy downcall
Quoted 38 lines · Simo Sorce
Quoted 37 lines · J. Bruce Fields
Quoted 35 lines · Simo Sorce
Quoted 21 lines · J. Bruce Fields
From: "J. Bruce Fields" <[email protected]>
For every other problem here we bail out with an error, but here for some reason we're setting a negative cache entry (with, note, an undefined expiry).
It seems simplest just to bail out in the same way as we do in other cases.
Cc Simo Sorce <[email protected]>Reported-by Andi Kleen <[email protected]>Signed-off-by J. Bruce Fields <[email protected]>--- net/sunrpc/auth_gss/svcauth_gss.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)net/sunrpc/auth_gss/svcauth_gss.c +0
-0net/sunrpc/auth_gss/svcauth_gss.c static int gss_proxy_save_rsc(struct cache_detail *cd, · line 1174
Quoted 10 lines
net/sunrpc/auth_gss/svcauth_gss.c +2
-2@@ -1167,8 +1167,8 @@ static int gss_proxy_save_rsc(struct cache_detail *cd,if (!ud->found_creds) {/* userspace seem buggy, we should always get at least a * mapping to nobody */ -dprintk ("RPC: No creds found, marking Negative!\n" ); -set_bit (CACHE_NEGATIVE , &rsci.h.flags); +dprintk ("RPC: No creds found!\n" ); + goto out; }else {/* steal creds */ IIRC, we are doing this to avoid rapid upcall loops in the kernel, where we keep hammering upcalls out and keep getting an error back.
Quoted 8 lines
Quoted 6 lines
Looks like returning an error instead results in closing the connection to the client, so, depends how the client replies I guess.
In any case I don't see why we'd treat this particular gss-proxy bug differently than we would any other (like, say, passing down bad xdr, or a gss context that we can't import).
Uhmm I did not recall that, I guess the change is ok then.
OK, thanks for the review.
--b.
PATCH 3/4 svcrpc: handle some gssproxy encoding errors
In reply to J. Bruce FieldsFrom: "J. Bruce Fields" <[email protected]>
--- net/sunrpc/auth_gss/gss_rpc_xdr.c | 29 ++++++++++++++--------------- 1 file changed, 14 insertions(+), 15 deletions(-)
net/sunrpc/auth_gss/gss_rpc_xdr.c +14 -15
@@ -559,6 +559,8 @@ static int gssx_enc_cred(struct xdr_stream *xdr,/* cred->elements */ err =dummy_enc_credel_array (xdr, &cred->elements); +if (err) +return err;/* cred->cred_handle_reference */ err =gssx_enc_buffer (xdr, &cred->cred_handle_reference);
@@ -740,22 +742,20 @@ void gssx_enc_accept_sec_context(struct rpc_rqst *req, goto done;/* arg->context_handle */ -if (arg->context_handle) { +if (arg->context_handle) err =gssx_enc_ctx (xdr, arg->context_handle); -if (err) - goto done; - }else { +else err =gssx_enc_bool (xdr,0 ); - } +if (err) + goto done;/* arg->cred_handle */ -if (arg->cred_handle) { +if (arg->cred_handle) err =gssx_enc_cred (xdr, arg->cred_handle); -if (err) - goto done; - }else { +else err =gssx_enc_bool (xdr,0 ); - } +if (err) + goto done;/* arg->input_token */ err =gssx_enc_in_token (xdr, &arg->input_token);
@@ -763,13 +763,12 @@ void gssx_enc_accept_sec_context(struct rpc_rqst *req, goto done;/* arg->input_cb */ -if (arg->input_cb) { +if (arg->input_cb) err =gssx_enc_cb (xdr, arg->input_cb); -if (err) - goto done; - }else { +else err =gssx_enc_bool (xdr,0 ); - } +if (err) + goto done; err =gssx_enc_bool (xdr, arg->ret_deleg_cred);if (err)
Signature · 1 line
1.7.9.5
PATCH 4/4 gss_krb5: document that we ignore sequence number
In reply to J. Bruce FieldsFrom: "J. Bruce Fields" <[email protected]>
A couple times recently somebody has noticed that we're ignoring a sequence number here and wondered whether there's a bug.
In fact, there's not. Thanks to Andy Adamson for pointing out a useful explanation in rfc 2203. Add comments citing that rfc, and remove "seqnum" to prevent static checkers complaining about unused variables.
--- net/sunrpc/auth_gss/gss_krb5_unseal.c | 8 ++++---- net/sunrpc/auth_gss/gss_krb5_wrap.c | 6 ++++-- 2 files changed, 8 insertions(+), 6 deletions(-)
net/sunrpc/auth_gss/gss_krb5_unseal.c +4 -4
@@ -150,7 +150,6 @@ gss_verify_mic_v2(struct krb5_ctx *ctx,struct xdr_netobj cksumobj = {.len =sizeof (cksumdata), .data = cksumdata};s32 now; -u64 seqnum;u8 *ptr = read_token->data;u8 *cksumkey;u8 flags;
@@ -197,9 +196,10 @@ gss_verify_mic_v2(struct krb5_ctx *ctx,if (now > ctx->endtime)return GSS_S_CONTEXT_EXPIRED ; -/* do sequencing checks */ - - seqnum =be64_to_cpup ((__be64 *)ptr +8 ); +/* +* NOTE: the sequence number at ptr + 8 is skipped, rpcsec_gss +* doesn't want it checked; see page 6 of rfc 2203. +*/ return GSS_S_COMPLETE ; }
net/sunrpc/auth_gss/gss_krb5_wrap.c +4 -2
@@ -489,7 +489,6 @@ static u32 gss_unwrap_kerberos_v2(struct krb5_ctx *kctx, int offset, struct xdr_buf *buf) { s32 now; -u64 seqnum;u8 *ptr;u8 flags =0x00 ;u16 ec, rrc;
@@ -525,7 +524,10 @@ gss_unwrap_kerberos_v2(struct krb5_ctx *kctx, int offset, struct xdr_buf *buf) ec =be16_to_cpup ((__be16 *)(ptr +4 )); rrc =be16_to_cpup ((__be16 *)(ptr +6 )); - seqnum =be64_to_cpup ((__be64 *)(ptr +8 )); +/* +* NOTE: the sequence number at ptr + 8 is skipped, rpcsec_gss +* doesn't want it checked; see page 6 of rfc 2203. +*/ if (rrc !=0 )rotate_left (offset +16 , buf, rrc);
Signature · 1 line
1.7.9.5