linux-kernel - Re: Re: [PATCH] RDMA/rxe: Fix mismatched max_msg

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <9e833b88-4efe-48eb-a605-984f5ab7f49f@bytedance.com>
Date: Tue, 17 Dec 2024 11:37:45 +0800
From: zhenwei pi <pizhenwei@...edance.com>
To: Zhu Yanjun <yanjun.zhu@...ux.dev>, jgg@...pe.ca
Cc: zyjzyj2000@...il.com, leon@...nel.org, linux-rdma@...r.kernel.org,
 linux-kernel@...r.kernel.org
Subject: Re: Re: [PATCH] RDMA/rxe: Fix mismatched max_msg_sz



On 12/17/24 00:47, Zhu Yanjun wrote:
> 在 2024/12/16 13:19, zhenwei pi 写道:
>> User mode queries max_msg_sz as 0x800000 by command 'ibv_devinfo -v',
>> however ibv_post_send/ibv_post_recv has a limit of 2^31. Fix this
>> mismatched information.
>>
> 
> This is a buf fix. Perhaps Fixes tag is needed?
> 

Hi,

Please amend this on applying patch:
Fixes: f605f26ea196 ("RDMA/rxe: Protect QP state with qp->state_lock")
Fixes: 5bf944f24129 ("RDMA/rxe: Add error messages")

If v2 is needed, please let me know.

>> Signed-off-by: zhenwei pi <pizhenwei@...edance.com>
>> ---
>>   drivers/infiniband/sw/rxe/rxe_param.h | 2 +-
>>   drivers/infiniband/sw/rxe/rxe_verbs.c | 5 ++---
>>   2 files changed, 3 insertions(+), 4 deletions(-)
>>
>> diff --git a/drivers/infiniband/sw/rxe/rxe_param.h b/drivers/ 
>> infiniband/sw/rxe/rxe_param.h
>> index d2f57ead78ad..003f681e5dc0 100644
>> --- a/drivers/infiniband/sw/rxe/rxe_param.h
>> +++ b/drivers/infiniband/sw/rxe/rxe_param.h
>> @@ -129,7 +129,7 @@ enum rxe_device_param {
>>   enum rxe_port_param {
>>       RXE_PORT_GID_TBL_LEN        = 1024,
>>       RXE_PORT_PORT_CAP_FLAGS        = IB_PORT_CM_SUP,
>> -    RXE_PORT_MAX_MSG_SZ        = 0x800000,
>> +    RXE_PORT_MAX_MSG_SZ        = (1UL << 31),
> 
> In the above, perhaps the parentheses is not necessary.
> Except that, I am fine with this.
> 
> Thanks a lot.
> Review-by: Zhu Yanjun <yanjun.zhu@...ux.dev>
> 
> Zhu Yanjun
> 
>>       RXE_PORT_BAD_PKEY_CNTR        = 0,
>>       RXE_PORT_QKEY_VIOL_CNTR        = 0,
>>       RXE_PORT_LID            = 0,
>> diff --git a/drivers/infiniband/sw/rxe/rxe_verbs.c b/drivers/ 
>> infiniband/sw/rxe/rxe_verbs.c
>> index 5c18f7e342f2..ffd5b07ad3e6 100644
>> --- a/drivers/infiniband/sw/rxe/rxe_verbs.c
>> +++ b/drivers/infiniband/sw/rxe/rxe_verbs.c
>> @@ -688,7 +688,7 @@ static int validate_send_wr(struct rxe_qp *qp, 
>> const struct ib_send_wr *ibwr,
>>           for (i = 0; i < ibwr->num_sge; i++)
>>               length += ibwr->sg_list[i].length;
>> -        if (length > (1UL << 31)) {
>> +        if (length > RXE_PORT_MAX_MSG_SZ) {
>>               rxe_err_qp(qp, "message length too long\n");
>>               break;
>>           }
>> @@ -972,8 +972,7 @@ static int post_one_recv(struct rxe_rq *rq, const 
>> struct ib_recv_wr *ibwr)
>>       for (i = 0; i < num_sge; i++)
>>           length += ibwr->sg_list[i].length;
>> -    /* IBA max message size is 2^31 */
>> -    if (length >= (1UL<<31)) {
>> +    if (length > RXE_PORT_MAX_MSG_SZ) {
>>           err = -EINVAL;
>>           rxe_dbg("message length too long\n");
>>           goto err_out;
>