RE: [PATCH 5/7] cifs: smbd: Protect memory registration using RCU

From: Long Li
Date: Tue May 08 2018 - 18:16:34 EST

Next message: Kent Overstreet: "[PATCH 0/2] bcachefs: on disk data structures, ioctl interface - review requested for upstreaming"
Previous message: Theodore Y. Ts'o: "Re: [Ksummit-discuss] bug-introducing patches"
In reply to: kbuild test robot: "Re: [PATCH 5/7] cifs: smbd: Protect memory registration using RCU"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Hi Steve

Please drop this patch. I will find another solution.

The other patches in the series can be reviewed and applied.

Long

> -----Original Message-----
> From: kbuild test robot <lkp@xxxxxxxxx>
> Sent: Monday, May 7, 2018 11:36 PM
> To: Long Li <longli@xxxxxxxxxxxxxxxxx>
> Cc: kbuild-all@xxxxxx; Steve French <sfrench@xxxxxxxxx>; linux-
> cifs@xxxxxxxxxxxxxxx; samba-technical@xxxxxxxxxxxxxxx; linux-
> kernel@xxxxxxxxxxxxxxx; linux-rdma@xxxxxxxxxxxxxxx; Long Li
> <longli@xxxxxxxxxxxxx>
> Subject: Re: [PATCH 5/7] cifs: smbd: Protect memory registration using RCU
>
> Hi Long,
>
> Thank you for the patch! Perhaps something to improve:
>
> [auto build test WARNING on cifs/for-next] [also build test WARNING on
> v4.17-rc4 next-20180507] [if your patch is applied to the wrong git tree,
> please drop us a note to help improve the system]
>
> url:
> https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub
> .com%2F0day-ci%2Flinux%2Fcommits%2FLong-Li%2Fcifs-smbd-Make-upper-
> layer-decide-when-to-destroy-the-transport%2F20180508-
> 110150&data=02%7C01%7Clongli%40microsoft.com%7C56045b01aeae4e53b1
> 9408d5b4ae0502%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C6366
> 13581873868980&sdata=a1Ds3qI7sRrLCE77Eoasifonm1rUQYi7rdUEnLnGOB4%
> 3D&reserved=0
> base: git://git.samba.org/sfrench/cifs-2.6.git for-next
> reproduce:
> # apt-get install sparse
> make ARCH=x86_64 allmodconfig
> make C=1 CF=-D__CHECK_ENDIAN__
>
>
> sparse warnings: (new ones prefixed by >>)
>
> fs/cifs/smb2pdu.c:110:47: sparse: expression using sizeof(void)
> fs/cifs/smb2pdu.c:681:26: sparse: expression using sizeof(void)
> >> fs/cifs/smb2pdu.c:2613:17: sparse: incompatible types in comparison
> >> expression (different address spaces)
> fs/cifs/smb2pdu.c:2993:17: sparse: incompatible types in comparison
> expression (different address spaces)
> fs/cifs/smb2pdu.c:3251:23: sparse: expression using sizeof(void)
> fs/cifs/smb2pdu.c:3251:23: sparse: expression using sizeof(void)
> fs/cifs/smb2pdu.c:3252:23: sparse: expression using sizeof(void)
> fs/cifs/smb2pdu.c:3713:17: sparse: expression using sizeof(void)
> fs/cifs/smb2pdu.c:3713:17: sparse: expression using sizeof(void)
>
> vim +2613 fs/cifs/smb2pdu.c
>
> 2565
> 2566 /*
> 2567 * To form a chain of read requests, any read requests after the first
> should
> 2568 * have the end_of_chain boolean set to true.
> 2569 */
> 2570 static int
> 2571 smb2_new_read_req(void **buf, unsigned int *total_len,
> 2572 struct cifs_io_parms *io_parms, struct cifs_readdata *rdata,
> 2573 unsigned int remaining_bytes, int request_type)
> 2574 {
> 2575 int rc = -EACCES;
> 2576 struct smb2_read_plain_req *req = NULL;
> 2577 struct smb2_sync_hdr *shdr;
> 2578 struct TCP_Server_Info *server;
> 2579
> 2580 rc = smb2_plain_req_init(SMB2_READ, io_parms->tcon, (void
> **) &req,
> 2581 total_len);
> 2582 if (rc)
> 2583 return rc;
> 2584
> 2585 server = io_parms->tcon->ses->server;
> 2586 if (server == NULL)
> 2587 return -ECONNABORTED;
> 2588
> 2589 shdr = &req->sync_hdr;
> 2590 shdr->ProcessId = cpu_to_le32(io_parms->pid);
> 2591
> 2592 req->PersistentFileId = io_parms->persistent_fid;
> 2593 req->VolatileFileId = io_parms->volatile_fid;
> 2594 req->ReadChannelInfoOffset = 0; /* reserved */
> 2595 req->ReadChannelInfoLength = 0; /* reserved */
> 2596 req->Channel = 0; /* reserved */
> 2597 req->MinimumCount = 0;
> 2598 req->Length = cpu_to_le32(io_parms->length);
> 2599 req->Offset = cpu_to_le64(io_parms->offset);
> 2600 #ifdef CONFIG_CIFS_SMB_DIRECT
> 2601 /*
> 2602 * If we want to do a RDMA write, fill in and append
> 2603 * smbd_buffer_descriptor_v1 to the end of read request
> 2604 */
> 2605 if (server->rdma && rdata && !server->sign &&
> 2606 rdata->bytes >= server->smbd_conn-
> >rdma_readwrite_threshold) {
> 2607
> 2608 struct smbd_buffer_descriptor_v1 *v1;
> 2609 bool need_invalidate =
> 2610 io_parms->tcon->ses->server->dialect ==
> SMB30_PROT_ID;
> 2611
> 2612 rcu_read_lock();
> > 2613 rcu_dereference(server->smbd_conn);
> 2614 rdata->mr = smbd_register_mr(
> 2615 server->smbd_conn, rdata->pages,
> 2616 rdata->nr_pages, rdata->tailsz,
> 2617 true, need_invalidate);
> 2618 rcu_read_unlock();
> 2619 if (!rdata->mr)
> 2620 return -ENOBUFS;
> 2621
> 2622 req->Channel =
> SMB2_CHANNEL_RDMA_V1_INVALIDATE;
> 2623 if (need_invalidate)
> 2624 req->Channel = SMB2_CHANNEL_RDMA_V1;
> 2625 req->ReadChannelInfoOffset =
> 2626 cpu_to_le16(offsetof(struct
> smb2_read_plain_req, Buffer));
> 2627 req->ReadChannelInfoLength =
> 2628 cpu_to_le16(sizeof(struct
> smbd_buffer_descriptor_v1));
> 2629 v1 = (struct smbd_buffer_descriptor_v1 *) &req-
> >Buffer[0];
> 2630 v1->offset = cpu_to_le64(rdata->mr->mr->iova);
> 2631 v1->token = cpu_to_le32(rdata->mr->mr->rkey);
> 2632 v1->length = cpu_to_le32(rdata->mr->mr->length);
> 2633
> 2634 *total_len += sizeof(*v1) - 1;
> 2635 }
> 2636 #endif
> 2637 if (request_type & CHAINED_REQUEST) {
> 2638 if (!(request_type & END_OF_CHAIN)) {
> 2639 /* next 8-byte aligned request */
> 2640 *total_len = DIV_ROUND_UP(*total_len, 8) *
> 8;
> 2641 shdr->NextCommand =
> cpu_to_le32(*total_len);
> 2642 } else /* END_OF_CHAIN */
> 2643 shdr->NextCommand = 0;
> 2644 if (request_type & RELATED_REQUEST) {
> 2645 shdr->Flags |=
> SMB2_FLAGS_RELATED_OPERATIONS;
> 2646 /*
> 2647 * Related requests use info from previous
> read request
> 2648 * in chain.
> 2649 */
> 2650 shdr->SessionId = 0xFFFFFFFF;
> 2651 shdr->TreeId = 0xFFFFFFFF;
> 2652 req->PersistentFileId = 0xFFFFFFFF;
> 2653 req->VolatileFileId = 0xFFFFFFFF;
> 2654 }
> 2655 }
> 2656 if (remaining_bytes > io_parms->length)
> 2657 req->RemainingBytes =
> cpu_to_le32(remaining_bytes);
> 2658 else
> 2659 req->RemainingBytes = 0;
> 2660
> 2661 *buf = req;
> 2662 return rc;
> 2663 }
> 2664
>
> ---
> 0-DAY kernel test infrastructure Open Source Technology Center
> https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.01
> .org%2Fpipermail%2Fkbuild-
> all&data=02%7C01%7Clongli%40microsoft.com%7C56045b01aeae4e53b19408
> d5b4ae0502%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C63661358
> 1873868980&sdata=y3BYFeKEayhIfrs6hodamCPCyz8n3YQdN1yX3yXKs80%3D
> &reserved=0 Intel Corporation

Next message: Kent Overstreet: "[PATCH 0/2] bcachefs: on disk data structures, ioctl interface - review requested for upstreaming"
Previous message: Theodore Y. Ts'o: "Re: [Ksummit-discuss] bug-introducing patches"
In reply to: kbuild test robot: "Re: [PATCH 5/7] cifs: smbd: Protect memory registration using RCU"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]