Received-SPF: pass (google.com: domain of linux-kernel+bounces-46854-linux.lists.archive=gmail.com@vger.kernel.org designates 139.178.88.99 as permitted sender) client-ip=139.178.88.99;
Message-ID: <e89dc56f-556e-4743-8bd0-4050e5829de1@cornelisnetworks.com>
Date: Wed, 31 Jan 2024 12:01:54 -0500
User-Agent: Mozilla Thunderbird
Subject: Re: [PATCH] IB/hfi1: Fix sdma.h tx->num_descs off-by-one error (take
 two)
Content-Language: en-US
To: Leon Romanovsky <leon@kernel.org>, Daniel Vacek <neelx@redhat.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>,
 Brendan Cunningham <bcunningham@cornelisnetworks.com>,
 Patrick Kelsey <pat.kelsey@cornelisnetworks.com>, stable@vger.kernel.org,
 Mats Kronberg <kronberg@nsc.liu.se>, linux-rdma@vger.kernel.org,
 linux-kernel@vger.kernel.org
References: <20240126152125.869509-1-neelx@redhat.com>
 <20240131125037.GF71813@unreal>
From: Dennis Dalessandro <dennis.dalessandro@cornelisnetworks.com>
In-Reply-To: <20240131125037.GF71813@unreal>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk
MIME-Version: 1.0
X-MS-Exchange-AntiSpam-MessageData-ChunkCount: 1
X-MS-Exchange-AntiSpam-MessageData-0:
	=?utf-8?B?anozK0ZWekNlVzBXZ0tVZFJiSktBWmpiQWwrNENEN3RWckJhNWxQM3kwWG5M?=
 =?utf-8?B?MTQ2Y3hXNE9ZcHZKU3Jab2Uwd3diMlM4RUE1NS9qbisvYUJ4Wm5oKzVCUGho?=
 =?utf-8?B?TmFpU0FNSGcyaFltMlIrUXY2MlFvS0d6SWc3WWN0NGlGdUxKb2xRNGVpZ3dW?=
 =?utf-8?B?eFRUbkgraUVGOUhBMFJVWS8yVDA0RERXWEE3bjhpa09KVm84YnpTd3NNeTJZ?=
 =?utf-8?B?SFV1c045Y0JxRlREUUgzTHhLb2JIdVJqQU43MEI2YzV2ZmRJcUNjTW9yYU5P?=
 =?utf-8?B?U1VEUUdYVFNwSmNDbDBML3FDenZjZldTaFR6dFp6RXF5aTRoc0RneVJ4aFQ3?=
 =?utf-8?B?b1c4V1ZaTktJOEJ6b0lzT1Z1SHlpT2pucGl1djhycnhac0xYdmpLbHNwbGxz?=
 =?utf-8?B?cXg0a3VVMW1aTXY1eGNVQ1JibVRjbVhoaEp2RDNqejl1U0ZaZnFKL243bmhy?=
 =?utf-8?B?aGxNNC9GMTVycWFwUUlYc2pib1E2NTRJbFFlYk9ObnpzelhEZWpCYkJWZSs2?=
 =?utf-8?B?WjhvZVc0QURZcC9XRi9Ya3hITTlNNWhuK1g5ajhWMG00Z1JzZmhodHZOTWxk?=
 =?utf-8?B?eUo3aEFDa05CL3EyLzFYSnpsRUJTWm40SUZjMmZkeGVoOUtZSTNNWjlWa3R5?=
 =?utf-8?B?TE5aSzY4UFlTTWs5R3N5YnZybWxTQk1IWm8zeFVzc1lDWWtvVlZ5SUdvSFlp?=
 =?utf-8?B?cnEyMzllbnRUMTFOZ3BUaFZWRTV5a2J4eTNJZ09IdW82S1VNc2lkVUlXanVq?=
 =?utf-8?B?MVN4MHR0NE5uMnVESjJCQzNGRDA4WGR2ejVDNWhreGJvMGRjRXhyWitIcFow?=
 =?utf-8?B?VXpnZlZaUnNxMU1EOFdNMkhvNW5XOEt0UzJScjQzVG9HbEVaRWY1WUpBU2NJ?=
 =?utf-8?B?Sm8rU2pTTmVYNlNZalhScURHOGFTOTFUN0o2RlpyWkpCWWJwejZTN1lWenk5?=
 =?utf-8?B?SXY1NEdub0xiTGkvTE5LL0NoUWg0RlZLMGYxYTc5dTJpTlhlakxIN3RMTUtI?=
 =?utf-8?B?NUc5cmtiQWxBZmFlekV2WkhXWTdITEcyUUI1U1dJRDNvdXBRa3lpZDNYU1hL?=
 =?utf-8?B?VWpwT09udWNPVldiYmVhRVNzRTNXVW8rdzVsZnVsbEZvSFdsdFVUMXgyVlhp?=
 =?utf-8?B?K1BKZmRYMXhiS3l5WXltbklid05zLzI0NEFHbStEbGdZcVI2ZGk3Y0srMmhM?=
 =?utf-8?B?aGE0R2RWZTNJZ3lrbnlKTEhjU0VtdzZQelFYajMyZENvajI2MkQ1Y2Fhc3Yy?=
 =?utf-8?B?citoTFR6Rlc2a0l5RWh0WXFaSVVXajBxcjBsb3kvRHJNWGJuYTlKb0Z2VGR1?=
 =?utf-8?B?Y0lBWUtzWUhmckw5QThEZWZKdUpoWFprRzFERUtVVW1JakNPTU03dWhoSkFi?=
 =?utf-8?B?QmswVDVBSkdKL0RpZ1BsS2lBRnU1Rm4wUk9SVTZJcVZrekRoekxWTVcvWDdw?=
 =?utf-8?B?ZE5jSGdmYWJneTZIOUV3eG5BRC9KS1kzMWZOM3pFSDd6MDJTK0M0eTN2dXZP?=
 =?utf-8?B?UDhncEtUMkdsNyszSWZ5bEVqVkEzQ3lGa0hNb3lkdWoyditVTzdQOXR3Zktu?=
 =?utf-8?B?MnJ5ckpuSTJxczZkUnYrcDJCeFBQUXptYlAyMFRXVnFuQjcvNmdnYTdWQ0hJ?=
 =?utf-8?B?SW1Jd0lucFlHOFhzVlQ2Zm1aMGdKNWwvMjBlRFByWk1hMnkvWDJDQWFJZVBp?=
 =?utf-8?B?cjY4UlNpM0RCc3B0ZW5wWUJiN3NmbTJjc2pBSEZZYlRDM1ZsK3RGdHNRUzhp?=
 =?utf-8?B?TGswNjhMZXlIekFoYjNHUlBLS1Zpai9UeFRBcGdhOEdEUU9qdFNKejY3UVRF?=
 =?utf-8?B?SXZMcDRaTnN3NDE2dTVzcjJJZjJWMHpvZ1c5dE1WSjdXV3pWQTd4NUxRRDll?=
 =?utf-8?B?SHZmc2pyaEdmR1o0c1VMQnoxNDl2N21wN1NSSUFpOVpJcmRoTXFRVENnRmQ1?=
 =?utf-8?B?MThXZFFlSlp2VUdUS0RJL0QwVDh2MHZXL0lQSTE2aGtLcUQyNFdrRGd4bGZo?=
 =?utf-8?B?ZktxZGFLMlJURzlkUjB1c0JvTlAycStldExLOGNsQ2QrbGZmNG5zMlF3T25N?=
 =?utf-8?B?cUs1N2RTcXV6ZnRwT0pqQXRrcmw1SVBMMllTczI2a3AvYVZSTWpXNDFYVWR2?=
 =?utf-8?B?b2VjeVkzN3RabEJhSG8weTluc2dJalFQWWk0bTFRdEVPdmZmb08vKzAyMlQ2?=
 =?utf-8?Q?6VgfTMKAxOoHyu3pNDJEzSI=3D?=
X-OriginatorOrg: cornelisnetworks.com
X-MS-Exchange-CrossTenant-Network-Message-Id: 72858ef7-7852-46f9-0533-08dc227e564f
X-MS-Exchange-CrossTenant-AuthSource: BL0PR01MB4099.prod.exchangelabs.com
X-MS-Exchange-CrossTenant-AuthAs: Internal
X-MS-Exchange-CrossTenant-OriginalArrivalTime: 31 Jan 2024 17:01:59.0136
 (UTC)
X-MS-Exchange-CrossTenant-FromEntityHeader: Hosted
X-MS-Exchange-CrossTenant-Id: 4dbdb7da-74ee-4b45-8747-ef5ce5ebe68a
X-MS-Exchange-CrossTenant-MailboxType: HOSTED
X-MS-Exchange-CrossTenant-UserPrincipalName: a4pFXapGDFusocdnuf3dLxX/ceuiwjZTSl9T5SLwtLVJp4zfxkTM7XGKPmNj5ZUF9yKAK9JCh+xemVgvc7+Eo9MFA5Qe96gJP0zoDWDlpli4uH5mAGiXBqWU7vx4aVAf
X-MS-Exchange-Transport-CrossTenantHeadersStamped: SA0PR01MB6394

On 1/31/24 7:50 AM, Leon Romanovsky wrote:
> On Fri, Jan 26, 2024 at 04:21:23PM +0100, Daniel Vacek wrote:
>> Unfortunately the commit `fd8958efe877` introduced another error
>> causing the `descs` array to overflow. This reults in further crashes
>> easily reproducible by `sendmsg` system call.
>>
>> [ 1080.836473] general protection fault, probably for non-canonical address 0x400300015528b00a: 0000 [#1] PREEMPT SMP PTI
>> [ 1080.869326] RIP: 0010:hfi1_ipoib_build_ib_tx_headers.constprop.0+0xe1/0x2b0 [hfi1]
>> --
>> [ 1080.974535] Call Trace:
>> [ 1080.976990]  <TASK>
>> [ 1081.021929]  hfi1_ipoib_send_dma_common+0x7a/0x2e0 [hfi1]
>> [ 1081.027364]  hfi1_ipoib_send_dma_list+0x62/0x270 [hfi1]
>> [ 1081.032633]  hfi1_ipoib_send+0x112/0x300 [hfi1]
>> [ 1081.042001]  ipoib_start_xmit+0x2a9/0x2d0 [ib_ipoib]
>> [ 1081.046978]  dev_hard_start_xmit+0xc4/0x210
>> --
>> [ 1081.148347]  __sys_sendmsg+0x59/0xa0
>>
>> crash> ipoib_txreq 0xffff9cfeba229f00
>> struct ipoib_txreq {
>>   txreq = {
>>     list = {
>>       next = 0xffff9cfeba229f00,
>>       prev = 0xffff9cfeba229f00
>>     },
>>     descp = 0xffff9cfeba229f40,
>>     coalesce_buf = 0x0,
>>     wait = 0xffff9cfea4e69a48,
>>     complete = 0xffffffffc0fe0760 <hfi1_ipoib_sdma_complete>,
>>     packet_len = 0x46d,
>>     tlen = 0x0,
>>     num_desc = 0x0,
>>     desc_limit = 0x6,
>>     next_descq_idx = 0x45c,
>>     coalesce_idx = 0x0,
>>     flags = 0x0,
>>     descs = {{
>>         qw = {0x8024000120dffb00, 0x4}  # SDMA_DESC0_FIRST_DESC_FLAG (bit 63)
>>       }, {
>>         qw = {  0x3800014231b108, 0x4}
>>       }, {
>>         qw = { 0x310000e4ee0fcf0, 0x8}
>>       }, {
>>         qw = {  0x3000012e9f8000, 0x8}
>>       }, {
>>         qw = {  0x59000dfb9d0000, 0x8}
>>       }, {
>>         qw = {  0x78000e02e40000, 0x8}
>>       }}
>>   },
>>   sdma_hdr =  0x400300015528b000,  <<< invalid pointer in the tx request structure
>>   sdma_status = 0x0,                   SDMA_DESC0_LAST_DESC_FLAG (bit 62)
>>   complete = 0x0,
>>   priv = 0x0,
>>   txq = 0xffff9cfea4e69880,
>>   skb = 0xffff9d099809f400
>> }
>>
>> With this patch the crashes are no longer reproducible and the machine is stable.
>>
>> Note, the header file changes are just an unrelated clean-up while I was looking
>> around trying to find the bug.
>>
>> Fixes: fd8958efe877 ("IB/hfi1: Fix sdma.h tx->num_descs off-by-one errors")
>> Cc: stable@vger.kernel.org
>> Reported-by: Mats Kronberg <kronberg@nsc.liu.se>
>> Tested-by: Mats Kronberg <kronberg@nsc.liu.se>
>> Signed-off-by: Daniel Vacek <neelx@redhat.com>
>> ---
>>  drivers/infiniband/hw/hfi1/sdma.c |  2 +-
>>  drivers/infiniband/hw/hfi1/sdma.h | 17 +++++++----------
>>  2 files changed, 8 insertions(+), 11 deletions(-)
>>
>> diff --git a/drivers/infiniband/hw/hfi1/sdma.c b/drivers/infiniband/hw/hfi1/sdma.c
>> index 6e5ac2023328a..b67d23b1f2862 100644
>> --- a/drivers/infiniband/hw/hfi1/sdma.c
>> +++ b/drivers/infiniband/hw/hfi1/sdma.c
>> @@ -3158,7 +3158,7 @@ int _pad_sdma_tx_descs(struct hfi1_devdata *dd, struct sdma_txreq *tx)
>>  {
>>  	int rval = 0;
>>  
>> -	if ((unlikely(tx->num_desc + 1 == tx->desc_limit))) {
>> +	if ((unlikely(tx->num_desc == tx->desc_limit))) {
> 
> Maybe, Dennis?

I actually have a patch that does exactly this one line change queued up to send
out.

The commit message for our fix is:
    If an SDMA send consists of exactly 6 descriptors and requires dword
    padding (in the 7th descriptor), the sdma_txreq descriptor array
    is not properly expanded and the packet will overflow into the
    container structure. This results in a panic when the send completion
    runs. The exact panic varies depending on what elements of the
    container structure get corrupted. The fix is to use the correct
    expression in _pad_sdma_tx_descs() to test the need to expand the
    descriptor array.


>>  		rval = _extend_sdma_tx_descs(dd, tx);
>>  		if (rval) {
>>  			__sdma_txclean(dd, tx);
>> diff --git a/drivers/infiniband/hw/hfi1/sdma.h b/drivers/infiniband/hw/hfi1/sdma.h
>> index d77246b48434f..362815a8da267 100644
>> --- a/drivers/infiniband/hw/hfi1/sdma.h
>> +++ b/drivers/infiniband/hw/hfi1/sdma.h
>> @@ -639,13 +639,13 @@ static inline void sdma_txclean(struct hfi1_devdata *dd, struct sdma_txreq *tx)
>>  static inline void _sdma_close_tx(struct hfi1_devdata *dd,
>>  				  struct sdma_txreq *tx)
>>  {
>> -	u16 last_desc = tx->num_desc - 1;
>> +	struct sdma_desc *desc = &tx->descp[tx->num_desc - 1];
>>  
>> -	tx->descp[last_desc].qw[0] |= SDMA_DESC0_LAST_DESC_FLAG;
>> -	tx->descp[last_desc].qw[1] |= dd->default_desc1;
>> +	desc->qw[0] |= SDMA_DESC0_LAST_DESC_FLAG;
>> +	desc->qw[1] |= dd->default_desc1;
>>  	if (tx->flags & SDMA_TXREQ_F_URGENT)
>> -		tx->descp[last_desc].qw[1] |= (SDMA_DESC1_HEAD_TO_HOST_FLAG |
>> -					       SDMA_DESC1_INT_REQ_FLAG);
>> +		desc->qw[1] |= (SDMA_DESC1_HEAD_TO_HOST_FLAG |
>> +				SDMA_DESC1_INT_REQ_FLAG);
> 
> Unrelated change which doesn't change anything.

Please drop.

> 
>>  }
>>  
>>  static inline int _sdma_txadd_daddr(
>> @@ -670,13 +670,10 @@ static inline int _sdma_txadd_daddr(
>>  	tx->tlen -= len;
>>  	/* special cases for last */
>>  	if (!tx->tlen) {
>> -		if (tx->packet_len & (sizeof(u32) - 1)) {
>> +		if (tx->packet_len & (sizeof(u32) - 1))
>>  			rval = _pad_sdma_tx_descs(dd, tx);
>> -			if (rval)
>> -				return rval;
>> -		} else {
>> +		else
>>  			_sdma_close_tx(dd, tx);
>> -		}
> 
> Same as before, unrelated change.

Agree. Please drop.