Patchwork [net] rds: Fix NULL pointer dereference in __rds_rdma_map

login
register
mail settings
Submitter Håkon Bugge
Date Dec. 6, 2017, 4:18 p.m.
Message ID <20171206161828.8368-1-Haakon.Bugge@oracle.com>
Download mbox | patch
Permalink /patch/10096441/
State New
Headers show

Comments

Håkon Bugge - Dec. 6, 2017, 4:18 p.m.
This is a fix for syzkaller719569, where memory registration was
attempted without any underlying transport being loaded.

Analysis of the case reveals that it is the setsockopt() RDS_GET_MR
(2) and RDS_GET_MR_FOR_DEST (7) that are vulnerable.

Here is an example stack trace when the bug is hit:

BUG: unable to handle kernel NULL pointer dereference at 00000000000000c0
IP: __rds_rdma_map+0x36/0x440 [rds]
PGD 2f93d03067 P4D 2f93d03067 PUD 2f93d02067 PMD 0
Oops: 0000 [#1] SMP
Modules linked in: bridge stp llc tun rpcsec_gss_krb5 nfsv4
dns_resolver nfs fscache rds binfmt_misc sb_edac intel_powerclamp
coretemp kvm_intel kvm irqbypass crct10dif_pclmul c rc32_pclmul
ghash_clmulni_intel pcbc aesni_intel crypto_simd glue_helper cryptd
iTCO_wdt mei_me sg iTCO_vendor_support ipmi_si mei ipmi_devintf nfsd
shpchp pcspkr i2c_i801 ioatd ma ipmi_msghandler wmi lpc_ich mfd_core
auth_rpcgss nfs_acl lockd grace sunrpc ip_tables ext4 mbcache jbd2
mgag200 i2c_algo_bit drm_kms_helper ixgbe syscopyarea ahci sysfillrect
sysimgblt libahci mdio fb_sys_fops ttm ptp libata sd_mod mlx4_core drm
crc32c_intel pps_core megaraid_sas i2c_core dca dm_mirror
dm_region_hash dm_log dm_mod
CPU: 48 PID: 45787 Comm: repro_set2 Not tainted 4.14.2-3.el7uek.x86_64 #2
Hardware name: Oracle Corporation ORACLE SERVER X5-2L/ASM,MOBO TRAY,2U, BIOS 31110000 03/03/2017
task: ffff882f9190db00 task.stack: ffffc9002b994000
RIP: 0010:__rds_rdma_map+0x36/0x440 [rds]
RSP: 0018:ffffc9002b997df0 EFLAGS: 00010202
RAX: 0000000000000000 RBX: ffff882fa2182580 RCX: 0000000000000000
RDX: 0000000000000000 RSI: ffffc9002b997e40 RDI: ffff882fa2182580
RBP: ffffc9002b997e30 R08: 0000000000000000 R09: 0000000000000002
R10: ffff885fb29e3838 R11: 0000000000000000 R12: ffff882fa2182580
R13: ffff882fa2182580 R14: 0000000000000002 R15: 0000000020000ffc
FS:  00007fbffa20b700(0000) GS:ffff882fbfb80000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00000000000000c0 CR3: 0000002f98a66006 CR4: 00000000001606e0
Call Trace:
 rds_get_mr+0x56/0x80 [rds]
 rds_setsockopt+0x172/0x340 [rds]
 ? __fget_light+0x25/0x60
 ? __fdget+0x13/0x20
 SyS_setsockopt+0x80/0xe0
 do_syscall_64+0x67/0x1b0
 entry_SYSCALL64_slow_path+0x25/0x25
RIP: 0033:0x7fbff9b117f9
RSP: 002b:00007fbffa20aed8 EFLAGS: 00000293 ORIG_RAX: 0000000000000036
RAX: ffffffffffffffda RBX: 00000000000c84a4 RCX: 00007fbff9b117f9
RDX: 0000000000000002 RSI: 0000400000000114 RDI: 000000000000109b
RBP: 00007fbffa20af10 R08: 0000000000000020 R09: 00007fbff9dd7860
R10: 0000000020000ffc R11: 0000000000000293 R12: 0000000000000000
R13: 00007fbffa20b9c0 R14: 00007fbffa20b700 R15: 0000000000000021

Code: 41 56 41 55 49 89 fd 41 54 53 48 83 ec 18 8b 87 f0 02 00 00 48
89 55 d0 48 89 4d c8 85 c0 0f 84 2d 03 00 00 48 8b 87 00 03 00 00 <48>
83 b8 c0 00 00 00 00 0f 84 25 03 00 0 0 48 8b 06 48 8b 56 08

The fix is to check the existence of an underlying transport in
__rds_rdma_map().

Signed-off-by: Håkon Bugge <haakon.bugge@oracle.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
---
 net/rds/rdma.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
santosh.shilimkar@oralce.com - Dec. 6, 2017, 4:34 p.m.
On 12/6/2017 8:18 AM, Håkon Bugge wrote:
> This is a fix for syzkaller719569, where memory registration was
> attempted without any underlying transport being loaded.
> 
> Analysis of the case reveals that it is the setsockopt() RDS_GET_MR
> (2) and RDS_GET_MR_FOR_DEST (7) that are vulnerable.
> 
> Here is an example stack trace when the bug is hit:
> 
> BUG: unable to handle kernel NULL pointer dereference at 00000000000000c0
> IP: __rds_rdma_map+0x36/0x440 [rds]
> PGD 2f93d03067 P4D 2f93d03067 PUD 2f93d02067 PMD 0
> Oops: 0000 [#1] SMP
> Modules linked in: bridge stp llc tun rpcsec_gss_krb5 nfsv4
> dns_resolver nfs fscache rds binfmt_misc sb_edac intel_powerclamp
> coretemp kvm_intel kvm irqbypass crct10dif_pclmul c rc32_pclmul
> ghash_clmulni_intel pcbc aesni_intel crypto_simd glue_helper cryptd
> iTCO_wdt mei_me sg iTCO_vendor_support ipmi_si mei ipmi_devintf nfsd
> shpchp pcspkr i2c_i801 ioatd ma ipmi_msghandler wmi lpc_ich mfd_core
> auth_rpcgss nfs_acl lockd grace sunrpc ip_tables ext4 mbcache jbd2
> mgag200 i2c_algo_bit drm_kms_helper ixgbe syscopyarea ahci sysfillrect
> sysimgblt libahci mdio fb_sys_fops ttm ptp libata sd_mod mlx4_core drm
> crc32c_intel pps_core megaraid_sas i2c_core dca dm_mirror
> dm_region_hash dm_log dm_mod
> CPU: 48 PID: 45787 Comm: repro_set2 Not tainted 4.14.2-3.el7uek.x86_64 #2
> Hardware name: Oracle Corporation ORACLE SERVER X5-2L/ASM,MOBO TRAY,2U, BIOS 31110000 03/03/2017
> task: ffff882f9190db00 task.stack: ffffc9002b994000
> RIP: 0010:__rds_rdma_map+0x36/0x440 [rds]
> RSP: 0018:ffffc9002b997df0 EFLAGS: 00010202
> RAX: 0000000000000000 RBX: ffff882fa2182580 RCX: 0000000000000000
> RDX: 0000000000000000 RSI: ffffc9002b997e40 RDI: ffff882fa2182580
> RBP: ffffc9002b997e30 R08: 0000000000000000 R09: 0000000000000002
> R10: ffff885fb29e3838 R11: 0000000000000000 R12: ffff882fa2182580
> R13: ffff882fa2182580 R14: 0000000000000002 R15: 0000000020000ffc
> FS:  00007fbffa20b700(0000) GS:ffff882fbfb80000(0000) knlGS:0000000000000000
> CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> CR2: 00000000000000c0 CR3: 0000002f98a66006 CR4: 00000000001606e0
> Call Trace:
>   rds_get_mr+0x56/0x80 [rds]
>   rds_setsockopt+0x172/0x340 [rds]
>   ? __fget_light+0x25/0x60
>   ? __fdget+0x13/0x20
>   SyS_setsockopt+0x80/0xe0
>   do_syscall_64+0x67/0x1b0
>   entry_SYSCALL64_slow_path+0x25/0x25
> RIP: 0033:0x7fbff9b117f9
> RSP: 002b:00007fbffa20aed8 EFLAGS: 00000293 ORIG_RAX: 0000000000000036
> RAX: ffffffffffffffda RBX: 00000000000c84a4 RCX: 00007fbff9b117f9
> RDX: 0000000000000002 RSI: 0000400000000114 RDI: 000000000000109b
> RBP: 00007fbffa20af10 R08: 0000000000000020 R09: 00007fbff9dd7860
> R10: 0000000020000ffc R11: 0000000000000293 R12: 0000000000000000
> R13: 00007fbffa20b9c0 R14: 00007fbffa20b700 R15: 0000000000000021
> 
> Code: 41 56 41 55 49 89 fd 41 54 53 48 83 ec 18 8b 87 f0 02 00 00 48
> 89 55 d0 48 89 4d c8 85 c0 0f 84 2d 03 00 00 48 8b 87 00 03 00 00 <48>
> 83 b8 c0 00 00 00 00 0f 84 25 03 00 0 0 48 8b 06 48 8b 56 08
> 
> The fix is to check the existence of an underlying transport in
> __rds_rdma_map().
> 
> Signed-off-by: Håkon Bugge <haakon.bugge@oracle.com>
> Reported-by: syzbot <syzkaller@googlegroups.com>
> ---
Thanks Haakon for testing and posting out the discussed change
out.

Acked-by: Santosh Shilimkar <santosh.shilimkar@oracle.com>
David Miller - Dec. 6, 2017, 8:45 p.m.
From: Håkon Bugge <Haakon.Bugge@oracle.com>
Date: Wed,  6 Dec 2017 17:18:28 +0100

> This is a fix for syzkaller719569, where memory registration was
> attempted without any underlying transport being loaded.
> 
> Analysis of the case reveals that it is the setsockopt() RDS_GET_MR
> (2) and RDS_GET_MR_FOR_DEST (7) that are vulnerable.
> 
> Here is an example stack trace when the bug is hit:
 ...
> The fix is to check the existence of an underlying transport in
> __rds_rdma_map().
> 
> Signed-off-by: Håkon Bugge <haakon.bugge@oracle.com>
> Reported-by: syzbot <syzkaller@googlegroups.com>

Applied and queued up for -stable, thanks.

Patch

diff --git a/net/rds/rdma.c b/net/rds/rdma.c
index 8886f15abe90..bc2f1e0977d6 100644
--- a/net/rds/rdma.c
+++ b/net/rds/rdma.c
@@ -183,7 +183,7 @@  static int __rds_rdma_map(struct rds_sock *rs, struct rds_get_mr_args *args,
 	long i;
 	int ret;
 
-	if (rs->rs_bound_addr == 0) {
+	if (rs->rs_bound_addr == 0 || !rs->rs_transport) {
 		ret = -ENOTCONN; /* XXX not a great errno */
 		goto out;
 	}