net: tipc: fix possible infoleak in tipc_mon_rcv()
Subsystems (3)
PATCH v2 1/1 net: tipc: fix possible infoleak in tipc_mon_rcv()
dom_bef is use to cache current domain record only if current domain exists. But when current domain does not exist, dom_bef will still be used in mon_identify_lost_members. This may lead to an information leak.
Fix this by adding a memset before using dom_bef.
---
v2: remove redundant 'dom_bef.member_cnt = 0;'
net/tipc/monitor.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
net/tipc/monitor.c +1 -1
@@ -534,7 +534,7 @@ void tipc_mon_rcv(struct net *net, void *data, u16 dlen, u32 addr, state->peer_gen = new_gen;/* Cache current domain record for later use */ - dom_bef.member_cnt =0 ; +memset (&dom_bef,0 ,sizeof (dom_bef)); dom = peer->domain;if (dom)memcpy (&dom_bef, dom, dom->len);
Signature · 1 line
2.25.1
Quoted 37 lines
-----Original Message-----
From: Hangyu Hua <[email protected]>
Sent: Tuesday, June 28, 2022 3:31 PM
To: [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; Tung Quang Nguyen <[email protected]>
Cc: [email protected]; [email protected]; [email protected]; Hangyu Hua <[email protected]>
Subject: [PATCH v2] net: tipc: fix possible infoleak intipc_mon_rcv()
dom_befis use to cache current domain record only if current domain exists. But when current domain does not exist,dom_befwill still be used inmon_identify_lost_members. This may lead to an information leak.Fix this by adding a memset before using
dom_bef.Fixes 35c55c9877f8 ("tipc: add neighbor monitoring framework")Signed-off-by Hangyu Hua <[email protected]>---
v2: remove redundant 'dom_bef.member_cnt = 0;'
net/tipc/monitor.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)net/tipc/monitor.c +1
-2@@ -534,7 +534,7 @@ void tipc_mon_rcv(struct net *net, void *data, u16 dlen, u32 addr, state->peer_gen = new_gen;/* Cache current domain record for later use */ - dom_bef.member_cnt =0 ; +memset (&dom_bef,0 ,sizeof (dom_bef)); dom = peer->domain;if (dom)memcpy (&dom_bef, dom, dom->len); --
Quoted 1 line
2.25.1
Jakub Kicinski
maintainerQuoted 3 lines · On Tue, 28 Jun 2022 16:31:22 +0800 Hangyu Hua wrote:
dom_befis use to cache current domain record only if current domain exists. But when current domain does not exist,dom_befwill still be used inmon_identify_lost_members. This may lead to an information leak.
AFAICT applied_bef must be zero if peer->domain was 0, so I don't think mon_identify_lost_members() will do anything.
Quoted 4 lines
Fix this by adding a memset before using
dom_bef.Fixes 35c55c9877f8 ("tipc: add neighbor monitoring framework")Signed-off-by Hangyu Hua <[email protected]>
Quoted 8 lines · Jakub Kicinski
Quoted 3 lines · On Tue, 28 Jun 2022 16:31:22 +0800 Hangyu Hua wrote:
dom_befis use to cache current domain record only if current domain exists. But when current domain does not exist,dom_befwill still be used inmon_identify_lost_members. This may lead to an information leak.AFAICT
applied_befmust be zero ifpeer->domainwas 0, so I don't thinkmon_identify_lost_members()will do anything.
void tipc_mon_rcv(struct net *net, void *data, u16 dlen, u32 addr,
struct tipc_mon_state *state, int bearer_id)
{
...
if (!dom || (dom->len < new_dlen)) {
kfree(dom);
dom = kmalloc(new_dlen, GFP_ATOMIC); <--- [1]
peer->domain = dom;
if (!dom)
goto exit;
}...
}
peer->domain will be NULL when [1] fails. But there will not change peer->applied to 0. In this case, if tipc_mon_rcv is called again then an information leak will happen.
Thanks,
Hangyu.
Quoted 4 lines
Quoted 4 lines
Fix this by adding a memset before using
dom_bef.Fixes 35c55c9877f8 ("tipc: add neighbor monitoring framework")Signed-off-by Hangyu Hua <[email protected]>
Jakub Kicinski
maintainerQuoted 21 lines · On Thu, 30 Jun 2022 17:19:21 +0800 Hangyu Hua wrote:
Quoted 8 lines · Jakub Kicinski
Quoted 3 lines · On Tue, 28 Jun 2022 16:31:22 +0800 Hangyu Hua wrote:
dom_befis use to cache current domain record only if current domain exists. But when current domain does not exist,dom_befwill still be used inmon_identify_lost_members. This may lead to an information leak.AFAICT
applied_befmust be zero ifpeer->domainwas 0, so I don't thinkmon_identify_lost_members()will do anything.void tipc_mon_rcv(struct net *net, void *data, u16 dlen, u32 addr,
structtipc_mon_state*state, intbearer_id)
{
...if (!dom || (dom->len < new_dlen)) { kfree(dom); dom = kmalloc(new_dlen, GFP_ATOMIC); <--- [1] peer->domain = dom; if (!dom) goto exit; }
Quoted 6 lines
...
}
peer->domainwill be NULL when [1] fails. But there will not changepeer->appliedto 0. In this case, iftipc_mon_rcvis called again then an information leak will happen.
I see, good analysis! Jon, Xue - is there a reason domain gets wiped on memory allocation failure? I'd think we should leave the previous pointer in place instead of freeing it first.