Re: [net v2 1/1] ice: Fix crash by keep old cfg when update TCs more than queues

From: Anatolii Gerasymenko
Date: Wed Aug 17 2022 - 02:58:11 EST


On 16.08.2022 14:44, Ding Hui wrote:
> On 2022/8/16 17:13, Anatolii Gerasymenko wrote:
>> On 15.08.2022 03:18, Ding Hui wrote:
>>> There are problems if allocated queues less than Traffic Classes.
>>>
>>> Commit a632b2a4c920 ("ice: ethtool: Prohibit improper channel config
>>> for DCB") already disallow setting less queues than TCs.
>>>
>>> Another case is if we first set less queues, and later update more TCs
>>> config due to LLDP, ice_vsi_cfg_tc() will failed but left dirty
>>> num_txq/rxq and tc_cfg in vsi, that will cause invalid porinter access.
>>
>> Nice catch. Looks good to me.
>
> Thanks, I'll send v3 later, could I add Acked-by: tag too?

Please add Reviewed-by: Anatolii Gerasymenko <anatolii.gerasymenko@xxxxxxxxx>
Thanks.

>>  
>>> [   95.968089] ice 0000:3b:00.1: More TCs defined than queues/rings allocated.
>>> [   95.968092] ice 0000:3b:00.1: Trying to use more Rx queues (8), than were allocated (1)!
>>> [   95.968093] ice 0000:3b:00.1: Failed to config TC for VSI index: 0
>>> [   95.969621] general protection fault: 0000 [#1] SMP NOPTI
>>> [   95.969705] CPU: 1 PID: 58405 Comm: lldpad Kdump: loaded Tainted: G     U  W  O     --------- -t - 4.18.0 #1
>>> [   95.969867] Hardware name: O.E.M/BC11SPSCB10, BIOS 8.23 12/30/2021
>>> [   95.969992] RIP: 0010:devm_kmalloc+0xa/0x60
>>> [   95.970052] Code: 5c ff ff ff 31 c0 5b 5d 41 5c c3 b8 f4 ff ff ff eb f4 0f 1f 40 00 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 48 89 f8 89 d1 <8b> 97 60 02 00 00 48 8d 7e 18 48 39 f7 72 3f 55 89 ce 53 48 8b 4c
>>> [   95.970344] RSP: 0018:ffffc9003f553888 EFLAGS: 00010206
>>> [   95.970425] RAX: dead000000000200 RBX: ffffea003c425b00 RCX: 00000000006080c0
>>> [   95.970536] RDX: 00000000006080c0 RSI: 0000000000000200 RDI: dead000000000200
>>> [   95.970648] RBP: dead000000000200 R08: 00000000000463c0 R09: ffff888ffa900000
>>> [   95.970760] R10: 0000000000000000 R11: 0000000000000002 R12: ffff888ff6b40100
>>> [   95.970870] R13: ffff888ff6a55018 R14: 0000000000000000 R15: ffff888ff6a55460
>>> [   95.970981] FS:  00007f51b7d24700(0000) GS:ffff88903ee80000(0000) knlGS:0000000000000000
>>> [   95.971108] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
>>> [   95.971197] CR2: 00007fac5410d710 CR3: 0000000f2c1de002 CR4: 00000000007606e0
>>> [   95.971309] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
>>> [   95.971419] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
>>> [   95.971530] PKRU: 55555554
>>> [   95.971573] Call Trace:
>>> [   95.971622]  ice_setup_rx_ring+0x39/0x110 [ice]
>>> [   95.971695]  ice_vsi_setup_rx_rings+0x54/0x90 [ice]
>>> [   95.971774]  ice_vsi_open+0x25/0x120 [ice]
>>> [   95.971843]  ice_open_internal+0xb8/0x1f0 [ice]
>>> [   95.971919]  ice_ena_vsi+0x4f/0xd0 [ice]
>>> [   95.971987]  ice_dcb_ena_dis_vsi.constprop.5+0x29/0x90 [ice]
>>> [   95.972082]  ice_pf_dcb_cfg+0x29a/0x380 [ice]
>>> [   95.972154]  ice_dcbnl_setets+0x174/0x1b0 [ice]
>>> [   95.972220]  dcbnl_ieee_set+0x89/0x230
>>> [   95.972279]  ? dcbnl_ieee_del+0x150/0x150
>>> [   95.972341]  dcb_doit+0x124/0x1b0
>>> [   95.972392]  rtnetlink_rcv_msg+0x243/0x2f0
>>> [   95.972457]  ? dcb_doit+0x14d/0x1b0
>>> [   95.972510]  ? __kmalloc_node_track_caller+0x1d3/0x280
>>> [   95.972591]  ? rtnl_calcit.isra.31+0x100/0x100
>>> [   95.972661]  netlink_rcv_skb+0xcf/0xf0
>>> [   95.972720]  netlink_unicast+0x16d/0x220
>>> [   95.972781]  netlink_sendmsg+0x2ba/0x3a0
>>> [   95.975891]  sock_sendmsg+0x4c/0x50
>>> [   95.979032]  ___sys_sendmsg+0x2e4/0x300
>>> [   95.982147]  ? kmem_cache_alloc+0x13e/0x190
>>> [   95.985242]  ? __wake_up_common_lock+0x79/0x90
>>> [   95.988338]  ? __check_object_size+0xac/0x1b0
>>> [   95.991440]  ? _copy_to_user+0x22/0x30
>>> [   95.994539]  ? move_addr_to_user+0xbb/0xd0
>>> [   95.997619]  ? __sys_sendmsg+0x53/0x80
>>> [   96.000664]  __sys_sendmsg+0x53/0x80
>>> [   96.003747]  do_syscall_64+0x5b/0x1d0
>>> [   96.006862]  entry_SYSCALL_64_after_hwframe+0x65/0xca
>>>
>>> Only update num_txq/rxq when passed check, and restore tc_cfg if setup
>>> queue map failed.
>>>
>>> Signed-off-by: Ding Hui <dinghui@xxxxxxxxxxxxxx>
>>
>> Please, also add Fixes tag.
>>
>>> ---
>>>   drivers/net/ethernet/intel/ice/ice_lib.c | 42 +++++++++++++++---------
>>>   1 file changed, 26 insertions(+), 16 deletions(-)
>>>
>>> ---
>>> v1:
>>> https://patchwork.kernel.org/project/netdevbpf/patch/20220812123933.5481-1-dinghui@xxxxxxxxxxxxxx/
>>>
>>> v2:
>>>    rewrite subject
>>>    rebase to net
>>>
>>> diff --git a/drivers/net/ethernet/intel/ice/ice_lib.c b/drivers/net/ethernet/intel/ice/ice_lib.c
>>> index a830f7f9aed0..6e64cca30351 100644
>>> --- a/drivers/net/ethernet/intel/ice/ice_lib.c
>>> +++ b/drivers/net/ethernet/intel/ice/ice_lib.c
>>> @@ -914,7 +914,7 @@ static void ice_set_dflt_vsi_ctx(struct ice_hw *hw, struct ice_vsi_ctx *ctxt)
>>>    */
>>>   static int ice_vsi_setup_q_map(struct ice_vsi *vsi, struct ice_vsi_ctx *ctxt)
>>>   {
>>> -    u16 offset = 0, qmap = 0, tx_count = 0, pow = 0;
>>> +    u16 offset = 0, qmap = 0, tx_count = 0, rx_count = 0, pow = 0;
>>>       u16 num_txq_per_tc, num_rxq_per_tc;
>>>       u16 qcount_tx = vsi->alloc_txq;
>>>       u16 qcount_rx = vsi->alloc_rxq;
>>> @@ -981,23 +981,25 @@ static int ice_vsi_setup_q_map(struct ice_vsi *vsi, struct ice_vsi_ctx *ctxt)
>>>        * at least 1)
>>>        */
>>>       if (offset)
>>> -        vsi->num_rxq = offset;
>>> +        rx_count = offset;
>>>       else
>>> -        vsi->num_rxq = num_rxq_per_tc;
>>> +        rx_count = num_rxq_per_tc;
>>>   -    if (vsi->num_rxq > vsi->alloc_rxq) {
>>> +    if (rx_count > vsi->alloc_rxq) {
>>>           dev_err(ice_pf_to_dev(vsi->back), "Trying to use more Rx queues (%u), than were allocated (%u)!\n",
>>> -            vsi->num_rxq, vsi->alloc_rxq);
>>> +            rx_count, vsi->alloc_rxq);
>>>           return -EINVAL;
>>>       }
>>>   -    vsi->num_txq = tx_count;
>>> -    if (vsi->num_txq > vsi->alloc_txq) {
>>> +    if (tx_count > vsi->alloc_txq) {
>>>           dev_err(ice_pf_to_dev(vsi->back), "Trying to use more Tx queues (%u), than were allocated (%u)!\n",
>>> -            vsi->num_txq, vsi->alloc_txq);
>>> +            tx_count, vsi->alloc_txq);
>>>           return -EINVAL;
>>>       }
>>>   +    vsi->num_txq = tx_count;
>>> +    vsi->num_rxq = rx_count;
>>> +
>>>       if (vsi->type == ICE_VSI_VF && vsi->num_txq != vsi->num_rxq) {
>>>           dev_dbg(ice_pf_to_dev(vsi->back), "VF VSI should have same number of Tx and Rx queues. Hence making them equal\n");
>>>           /* since there is a chance that num_rxq could have been changed
>>> @@ -3492,6 +3494,7 @@ ice_vsi_setup_q_map_mqprio(struct ice_vsi *vsi, struct ice_vsi_ctx *ctxt,
>>>       int tc0_qcount = vsi->mqprio_qopt.qopt.count[0];
>>>       u8 netdev_tc = 0;
>>>       int i;
>>> +    u16 new_txq, new_rxq;
>>
>> Please follow the Reverse Christmas Tree (RCT) convention.
>>
>>>       vsi->tc_cfg.ena_tc = ena_tc ? ena_tc : 1;
>>>   @@ -3530,21 +3533,24 @@ ice_vsi_setup_q_map_mqprio(struct ice_vsi *vsi, struct ice_vsi_ctx *ctxt,
>>>           }
>>>       }
>>>   -    /* Set actual Tx/Rx queue pairs */
>>> -    vsi->num_txq = offset + qcount_tx;
>>> -    if (vsi->num_txq > vsi->alloc_txq) {
>>> +    new_txq = offset + qcount_tx;
>>> +    if (new_txq > vsi->alloc_txq) {
>>>           dev_err(ice_pf_to_dev(vsi->back), "Trying to use more Tx queues (%u), than were allocated (%u)!\n",
>>> -            vsi->num_txq, vsi->alloc_txq);
>>> +            new_txq, vsi->alloc_txq);
>>>           return -EINVAL;
>>>       }
>>>   -    vsi->num_rxq = offset + qcount_rx;
>>> -    if (vsi->num_rxq > vsi->alloc_rxq) {
>>> +    new_rxq = offset + qcount_rx;
>>> +    if (new_rxq > vsi->alloc_rxq) {
>>>           dev_err(ice_pf_to_dev(vsi->back), "Trying to use more Rx queues (%u), than were allocated (%u)!\n",
>>> -            vsi->num_rxq, vsi->alloc_rxq);
>>> +            new_rxq, vsi->alloc_rxq);
>>>           return -EINVAL;
>>>       }
>>>   +    /* Set actual Tx/Rx queue pairs */
>>> +    vsi->num_txq = new_txq;
>>> +    vsi->num_rxq = new_rxq;
>>> +
>>>       /* Setup queue TC[0].qmap for given VSI context */
>>>       ctxt->info.tc_mapping[0] = cpu_to_le16(qmap);
>>>       ctxt->info.q_mapping[0] = cpu_to_le16(vsi->rxq_map[0]);
>>> @@ -3580,6 +3586,7 @@ int ice_vsi_cfg_tc(struct ice_vsi *vsi, u8 ena_tc)
>>>       struct device *dev;
>>>       int i, ret = 0;
>>>       u8 num_tc = 0;
>>> +    struct ice_tc_cfg old_tc_cfg;
>>
>> RCT here also.
>>   
>>>       dev = ice_pf_to_dev(pf);
>>>       if (vsi->tc_cfg.ena_tc == ena_tc &&
>>> @@ -3600,6 +3607,7 @@ int ice_vsi_cfg_tc(struct ice_vsi *vsi, u8 ena_tc)
>>>               max_txqs[i] = vsi->num_txq;
>>>       }
>>>   +    memcpy(&old_tc_cfg, &vsi->tc_cfg, sizeof(old_tc_cfg));
>>>       vsi->tc_cfg.ena_tc = ena_tc;
>>>       vsi->tc_cfg.numtc = num_tc;
>>>   @@ -3616,8 +3624,10 @@ int ice_vsi_cfg_tc(struct ice_vsi *vsi, u8 ena_tc)
>>>       else
>>>           ret = ice_vsi_setup_q_map(vsi, ctx);
>>>   -    if (ret)
>>> +    if (ret) {
>>> +        memcpy(&vsi->tc_cfg, &old_tc_cfg, sizeof(vsi->tc_cfg));
>>>           goto out;
>>> +    }
>>>         /* must to indicate which section of VSI context are being modified */
>>>       ctx->info.valid_sections = cpu_to_le16(ICE_AQ_VSI_PROP_RXQ_MAP_VALID);
>
>