- 03 Nov, 2015 2 commits
-
-
Yaowei Bai authored
As new_valid_dev always returns 1, so !new_valid_dev check is not needed, remove it. Signed-off-by:
Yaowei Bai <bywxiaobai@163.com> Signed-off-by:
Trond Myklebust <trond.myklebust@primarydata.com>
-
Andrzej Hajda authored
Due to incorrect len type bc_send_request returned always zero. The problem has been detected using proposed semantic patch scripts/coccinelle/tests/assign_signed_to_unsigned.cocci [1]. [1]: http://permalink.gmane.org/gmane.linux.kernel/2046107Signed-off-by:
Andrzej Hajda <a.hajda@samsung.com> Signed-off-by:
-
- 02 Nov, 2015 20 commits
-
-
git://git.linux-nfs.org/projects/anna/nfs-rdmaTrond Myklebust authored
NFS: NFSoRDMA Client Side Changes In addition to a variety of bugfixes, these patches are mostly geared at enabling both swap and backchannel support to the NFS over RDMA client. Signed-off-by:Anna Schumake <Anna.Schumaker@Netapp.com>
-
Chuck Lever authored
Forechannel transports get their own "bc_up" method to create an endpoint for the backchannel service. Signed-off-by:
Chuck Lever <chuck.lever@oracle.com> [Anna Schumaker: Add forward declaration of struct net to xprt.h] Signed-off-by:
-
Trond Myklebust authored
For loosely coupled pNFS/flexfiles systems, there is often no advantage at all in going through the MDS for I/O, since the MDS is subject to the same limitations as all other clients when talking to DSes. If a DS is unresponsive, I/O through the MDS will fail. For such systems, the only scalable solution is to have the pNFS clients retry doing pNFS, and so the protocol now provides a flag that allows the pNFS server to signal this. If LAYOUTGET returns FF_FLAGS_NO_IO_THRU_MDS, then we should assume that the MDS wants the client to retry using these devices, even if they were previously marked as being unavailable. To do so, we add a helper, ff_layout_mark_devices_valid() that will be called from layoutget. Signed-off-by: -
Trond Myklebust authored
If the pNFS/flexfiles file is mirrored, and a read to one mirror fails, then we should bump the mirror index, so that we retry to a different mirror. Once we've iterated through all mirrors and all failed, we can return the layout and issue a new LAYOUTGET. Signed-off-by: -
Chuck Lever authored
Allow the use of other transport classes when handling a backward direction RPC call. Signed-off-by:
Sagi Grimberg <sagig@mellanox.com> Tested-By:
Devesh Sharma <devesh.sharma@avagotech.com> Signed-off-by:
-
Chuck Lever authored
On NFSv4.1 mount points, the Linux NFS client uses this transport endpoint to receive backward direction calls and route replies back to the NFSv4.1 server. Signed-off-by:
"J. Bruce Fields" <bfields@fieldses.org> Reviewed-by:
-
Chuck Lever authored
Introduce a code path in the rpcrdma_reply_handler() to catch incoming backward direction RPC calls and route them to the ULP's backchannel server. Signed-off-by:
-
Chuck Lever authored
Backward direction RPC replies are sent via the client transport's send_request method, the same way forward direction RPC calls are sent. Signed-off-by:
-
Chuck Lever authored
Pre-allocate extra send and receive Work Requests needed to handle backchannel receives and sends. The transport doesn't know how many extra WRs to pre-allocate until the xprt_setup_backchannel() call, but that's long after the WRs are allocated during forechannel setup. So, use a fixed value for now. Signed-off-by:
-
Chuck Lever authored
xprtrdma's backward direction send and receive buffers are the same size as the forechannel's inline threshold, and must be pre- registered. The consumer has no control over which receive buffer the adapter chooses to catch an incoming backwards-direction call. Any receive buffer can be used for either a forward reply or a backward call. Thus both types of RPC message must all be the same size. Signed-off-by:
-
Chuck Lever authored
xprt_{setup,destroy}_backchannel() won't be adequate for RPC/RMDA bi-direction. In particular, receive buffers have to be pre- registered and posted in order to receive incoming backchannel requests. Add a virtual function call to allow the insertion of appropriate backchannel setup and destruction methods for each transport. In addition, freeing a backchannel request is a little different for RPC/RDMA. Introduce an rpc_xprt_op to handle the difference. Signed-off-by: -
Chuck Lever authored
Now that RPC replies are processed in a workqueue, there's no need to disable IRQs when managing send and receive buffers. This saves noticeable overhead per RPC. Signed-off-by:
-
Chuck Lever authored
Clean up: The reply tasklet is no longer used. Signed-off-by:
-
Chuck Lever authored
The reply tasklet is fast, but it's single threaded. After reply traffic saturates a single CPU, there's no more reply processing capacity. Replace the tasklet with a workqueue to spread reply handling across all CPUs. This also moves RPC/RDMA reply handling out of the soft IRQ context and into a context that allows sleeps. Signed-off-by:
-
Chuck Lever authored
The rb_send_bufs and rb_recv_bufs arrays are used to implement a pair of stacks for keeping track of free rpcrdma_req and rpcrdma_rep structs. Replace those arrays with free lists. To allow more than 512 RPCs in-flight at once, each of these arrays would be larger than a page (assuming 8-byte addresses and 4KB pages). Allowing up to 64K in-flight RPCs (as TCP now does), each buffer array would have to be 128 pages. That's an order-6 allocation. (Not that we're going there.) A list is easier to expand dynamically. Instead of allocating a larger array of pointers and copying the existing pointers to the new array, simply append more buffers to each list. This also makes it simpler to manage receive buffers that might catch backwards-direction calls, or to post receive buffers in bulk to amortize the overhead of ib_post_recv. Signed-off-by:
-
Chuck Lever authored
Clean up: The error cases in rpcrdma_reply_handler() almost never execute. Ensure the compiler places them out of the hot path. No behavior change expected. Signed-off-by:
-
Chuck Lever authored
Commit 8301a2c0 ("xprtrdma: Limit work done by completion handler") was supposed to prevent xprtrdma's upcall handlers from starving other softIRQ work by letting them return to the provider before all CQEs have been polled. The logic assumes the provider will call the upcall handler again immediately if the CQ is re-armed while there are still queued CQEs. This assumption is invalid. The IBTA spec says that after a CQ is armed, the hardware must interrupt only when a new CQE is inserted. xprtrdma can't rely on the provider calling again, even though some providers do. Therefore, leaving CQEs on queue makes sense only when there is another mechanism that ensures all remaining CQEs are consumed in a timely fashion. xprtrdma does not have such a mechanism. If a CQE remains queued, the transport can wait forever to send the next RPC. Finally, move the wcs array back onto the stack to ensure that the poll array is always local to the CPU where the completion upcall is running. Fixes: 8301a2c0 ("xprtrdma: Limit work done by completion ...") Signed-off-by:
-
Chuck Lever authored
ib_req_notify_cq(IB_CQ_REPORT_MISSED_EVENTS) returns a positive value if WCs were added to a CQ after the last completion upcall but before the CQ has been re-armed. Commit 7f23f6f6 ("xprtrmda: Reduce lock contention in completion handlers") assumed that when ib_req_notify_cq() returned a positive RC, the CQ had also been successfully re-armed, making it safe to return control to the provider without losing any completion signals. That is an invalid assumption. Change both completion handlers to continue polling while ib_req_notify_cq() returns a positive value. Fixes: 7f23f6f6 ("xprtrmda: Reduce lock contention in ...") Signed-off-by:
-
Chuck Lever authored
After adding a swapfile on an NFS/RDMA mount and removing the normal swap partition, I was able to push the NFS client well into swap without any issue. I forgot to swapoff the NFS file before rebooting. This pinned the NFS mount and the IB core and provider, causing shutdown to hang. I think this is expected and safe behavior. Probably shutdown scripts should "swapoff -a" before unmounting any filesystems. Signed-off-by:
-
Steve Wise authored
Unsignaled send WRs can get flushed as part of normal unmount, so don't log them as warnings. Signed-off-by:
Steve Wise <swise@opengridcomputing.com> Signed-off-by:
-
- 21 Oct, 2015 14 commits
-
-
Trond Myklebust authored
* bugfixes: NFSv4.1/pnfs: Retry through MDS when getting bad length of data nfs/blocklayout: Fix bad using of page offset in bl_read_pagelist NFS: Return directly if encode_sessionid fail NFS: Fix bad checking of max taglen in callback request NFS: Fix bad defines of callback response maxsize NFS: Use NFS4_MAX_SESSIONID_LEN directly for decode/encode sessionid NFS: Remove unneeded NFS_DEBUG checking before define NFSDBG_FACILITY NFS: Remove the left function defines in callback.h NFS: Remove the left global variable nfs_callback_tcpport NFS: Get rid of the unneeded addr stored in callback arguments nfsroot: make nfsroot to accept the 1024 bytes long directory name
-
Kinglong Mee authored
If non rpc-based layout driver return bad length of data, nfs retries by calling rpc_restart_call_prepare() that cause an NULL reference panic. This patch lets nfs retry through MDS for non rpc-based layout driver return bad length of data. [13034.883329] BUG: unable to handle kernel NULL pointer dereference at (null) [13034.884902] IP: [<ffffffffa00db372>] rpc_restart_call_prepare+0x62/0x90 [sunrpc] [13034.886558] PGD 0 [13034.888126] Oops: 0000 [#1] KASAN [13034.889710] Modules linked in: blocklayoutdriver(OE) nfsv4(OE) nfs(OE) fscache(E) nfsd(OE) xfs libcrc32c coretemp btrfs crct10dif_pclmul crc32_pclmul crc32c_intel ghash_clmulni_intel ppdev vmw_balloon auth_rpcgss shpchp nfs_acl lockd vmw_vmci parport_pc xor raid6_pq grace parport sunrpc i2c_piix4 vmwgfx drm_kms_helper ttm drm mptspi e1000 serio_raw scsi_transport_spi mptscsih mptbase ata_generic pata_acpi [last unloaded: fscache] [13034.898260] CPU: 0 PID: 10112 Comm: kworker/0:1 Tainted: G OE 4.3.0-rc5+ #279 [13034.899932] Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 07/02/2015 [13034.903342] Workqueue: events bl_read_cleanup [blocklayoutdriver] [13034.905059] task: ffff88006a9148c0 ti: ffff880035e90000 task.ti: ffff880035e90000 [13034.906827] RIP: 0010:[<ffffffffa00db372>] [<ffffffffa00db372>] rpc_restart_call_prepare+0x62/0x90 [sunrpc] [13034.910522] RSP: 0018:ffff880035e97b58 EFLAGS: 00010282 [13034.912378] RAX: fffffbfff04a5a94 RBX: ffff880068fe4858 RCX: 0000000000000003 [13034.914339] RDX: dffffc0000000000 RSI: 0000000000000003 RDI: 0000000000000282 [13034.916236] RBP: ffff880035e97b68 R08: 0000000000000001 R09: 0000000000000001 [13034.918229] R10: 0000000000000000 R11: 0000000000000001 R12: 0000000000000000 [13034.920007] R13: ffff880068fe4858 R14: ffff880068fe4a60 R15: 0000000000001000 [13034.921845] FS: 0000000000000000(0000) GS:ffffffff82247000(0000) knlGS:0000000000000000 [13034.923645] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [13034.925525] CR2: 0000000000000000 CR3: 00000000063dd000 CR4: 00000000001406f0 [13034.932808] Stack: [13034.934813] ffff880068fe4780 0000000000001000 ffff880035e97ba8 ffffffffa08800d2 [13034.936675] ffffffffa088029d ffff880068fe4780 ffff880068fe4858 ffffffffa089c0a0 [13034.938593] ffff880068fe47e0 ffff88005d59faf0 ffff880035e97be0 ffffffffa087e08f [13034.940454] Call Trace: [13034.942388] [<ffffffffa08800d2>] nfs_readpage_result+0x112/0x200 [nfs] [13034.944317] [<ffffffffa088029d>] ? nfs_readpage_done+0xdd/0x160 [nfs] [13034.946267] [<ffffffffa087e08f>] nfs_pgio_result+0x9f/0x120 [nfs] [13034.948166] [<ffffffffa09266cc>] pnfs_ld_read_done+0x7c/0x1e0 [nfsv4] [13034.950247] [<ffffffffa03b07ee>] bl_read_cleanup+0x2e/0x60 [blocklayoutdriver] [13034.952156] [<ffffffff810ebf62>] process_one_work+0x412/0x870 [13034.954102] [<ffffffff810ebe84>] ? process_one_work+0x334/0x870 [13034.955949] [<ffffffff810ebb50>] ? queue_delayed_work_on+0x40/0x40 [13034.957985] [<ffffffff810ec441>] worker_thread+0x81/0x6a0 [13034.959817] [<ffffffff810ec3c0>] ? process_one_work+0x870/0x870 [13034.961785] [<ffffffff810f43bd>] kthread+0x17d/0x1a0 [13034.963544] [<ffffffff810f4240>] ? kthread_create_on_node+0x330/0x330 [13034.965479] [<ffffffff81100428>] ? finish_task_switch+0x88/0x220 [13034.967223] [<ffffffff810f4240>] ? kthread_create_on_node+0x330/0x330 [13034.968929] [<ffffffff81b6ae5f>] ret_from_fork+0x3f/0x70 [13034.970534] [<ffffffff810f4240>] ? kthread_create_on_node+0x330/0x330 [13034.972176] Code: c7 43 50 40 84 0d a0 e8 3d fe 1c e1 48 8d 7b 58 c7 83 e4 00 00 00 00 00 00 00 e8 ca fe 1c e1 4c 8b 63 58 4c 89 e7 e8 be fe 1c e1 <49> 83 3c 24 00 74 12 48 c7 43 50 f0 a2 0e a0 b8 01 00 00 00 5b [13034.977148] RIP [<ffffffffa00db372>] rpc_restart_call_prepare+0x62/0x90 [sunrpc] [13034.978780] RSP <ffff880035e97b58> [13034.980399] CR2: 0000000000000000 Signed-off-by:
Kinglong Mee <kinglongmee@gmail.com> Signed-off-by:
-
Kinglong Mee authored
Blocklayout uses file offset for the read-back page's offset of first writing, it's definitely wrong, it writes data to bad address of page that cause userspace application segment fault. It must be the page base stored in header->args.pgbase. Also, the pg_offset has no influence with isect and extent length. Note: The offset of the non-first page is always zero. Ps: A test program will segment fault at read() as, #define _GNU_SOURCE #include <stdio.h> #include <stdlib.h> #include <sys/types.h> #include <sys/stat.h> #include <unistd.h> #include <fcntl.h> #include <errno.h> int main(int argc, char **argv) { char buf[2049]; char *filename = NULL; int fd = -1; if (argc < 2) { printf("Usage: %s filename\n", argv[0]); return 0; } filename = argv[1]; fd = open(filename, O_RDONLY | O_DIRECT); if (fd < 0) { printf("Open %s fail: %m\n", filename); return 1; } lseek(fd, 2048, SEEK_SET); if (read(fd, buf, sizeof(buf) - 1) != (sizeof(buf) - 1)) printf("Read 4096 bityes data from %s fail: %m\n", filename); out: close(fd); return 0; } Signed-off-by: -
Kinglong Mee authored
encode_sessionid() may return error, nfs needs process the return value. Signed-off-by:
-
Kinglong Mee authored
The taglen should be checked with CB_OP_TAGLEN_MAXSZ directly. Signed-off-by:
-
Kinglong Mee authored
As CB_OP_TAGLEN_MAXSZ, all XXX_MAXSZ should be defined as bit. Each operation should not cantains CB_OP_TAGLEN_MAXSZ. Signed-off-by:
-
Kinglong Mee authored
It's no need to define a temporary variables for NFS4_MAX_SESSIONID_LEN. Signed-off-by:
-
Kinglong Mee authored
It's not needed to checking NFS_DEBUG before define NFSDBG_FACILITY, remove it. Signed-off-by:
-
Kinglong Mee authored
Commit 778be232 "NFS do not find client in NFSv4 pg_authenticate" has remove the define and using of nfs4_set_callback_sessionid(), and commit 36281caa "NFSv4: Further clean-ups of delegation stateid validation" has update the checking of stateid, and move the code to nfs4proc.c. This patch remove those function defines left in callback.h Signed-off-by:
-
Kinglong Mee authored
Commit bbe0a3aa "NFS: make nfs_callback_tcpport per network context" has make nfs_callback_tcpport per network, but left the global nfs_callback_tcpport, remove it. Signed-off-by:
-
Kinglong Mee authored
Commit c36fca52 "NFS refactor nfs_find_client and reference client across callback processing" has store clp in cb_process_state which is set in cb_sequence. So that, it's unneeded to store address pointer in any callback arguments. Signed-off-by:
-
Li RongQing authored
although NFS_MAXPATHLEN is defined to 1024, nfs client hopes to accept a 1024 byte path, but nfs_root_parms is limited to 256, and the nfs path will truncated when a user inputs nfs path from kernel cmdline enlarge nfs_root_parms to 1024, to make it accept the 1024 bytes long directory name, since nfs_root_parms is defined as _initdata, it will be released after system bootup Signed-off-by:
Li RongQing <roy.qing.li@gmail.com> Signed-off-by:
-
Trond Myklebust authored
* nfsclone: nfs: add missing linux/types.h NFS: Fix an 'unused variable' complaint when #ifndef CONFIG_NFS_V4_2 nfs42: add NFS_IOC_CLONE_RANGE ioctl nfs42: respect clone_blksize nfs: get clone_blksize when probing fsinfo nfs42: add NFS_IOC_CLONE ioctl nfs42: add CLONE proc functions nfs42: add CLONE xdr functions
-
Peng Tao authored
After merging the nfs tree, today's linux-next build (powerpc allyesconfig produced this warning: ./usr/include/linux/nfs.h:40: found __[us]{8,16,32,64} type without #include <linux/types.h> Reported-by:Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by:
Peng Tao <tao.peng@primarydata.com> Signed-off-by:
-
- 18 Oct, 2015 3 commits
-
-
Linus Torvalds authored
-
git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linuxLinus Torvalds authored
Pull i2c fixes from Wolfram Sang: "Here are some bugfixes for the I2C subsystem. Kieran found a flaw in the recently renewed wake irq handling. Mika handled a user bug report where the ACPI info turned out to be unusable. I updated MAINTAINERS so that such bug reports will sooner get to the right people. Geert pointed me to a problem of some i2c drivers regarding PM which I fixed" * 'i2c/for-current' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux: i2c: designware: Do not use parameters from ACPI on Dell Inspiron 7348 MAINTAINERS: add maintainers for Synopsis Designware I2C drivers i2c: designware-platdrv: enable RuntimePM before registering to the core i2c: s3c2410: enable RuntimePM before registering to the core i2c: rcar: enable RuntimePM before registering to the core i2c: return probe deferred status on dev_pm_domain_attach
-
Mika Westerberg authored
ACPI SSCN/FMCN methods were originally added because then the platform can provide the most accurate HCNT/LCNT values to the driver. However, this seems not to be true for Dell Inspiron 7348 where using these causes the touchpad to fail in boot: i2c_hid i2c-DLL0675:00: failed to retrieve report from device. i2c_designware INT3433:00: i2c_dw_handle_tx_abort: lost arbitration i2c_hid i2c-DLL0675:00: failed to retrieve report from device. i2c_designware INT3433:00: controller timed out The values received from ACPI are (in fast mode): HCNT: 72 LCNT: 160 this translates to following timings (input clock is 100MHz on Broadwell): tHIGH: 720 ns (spec min 600 ns) tLOW: 1600 ns (spec min 1300 ns) Bus period: 2920 ns (assuming 300 ns tf and tr) Bus speed: 342.5 kHz Both tHIGH and tLOW are within the I2C specification. The calculated values when ACPI parameters are not used are (in fast mode): HCNT: 87 LCNT: 159 which translates to: tHIGH: 870 ns (spec min 600 ns) tLOW: 1590 ns (spec min 1300 ns) Bus period 3060 ns (assuming 300 ns tf and tr) Bus speed 326.8 kHz These values are also within the I2C specification. Since both ACPI and calculated values meet the I2C specification timing requirements it is hard to say why the touchpad does not function properly with the ACPI values except that the bus speed is higher in this case (but still well below the max 400kHz). Solve this by adding DMI quirk to the driver that disables using ACPI parameters on this particulare machine. Reported-by:
Pavel Roskin <plroskin@gmail.com> Signed-off-by:
Mika Westerberg <mika.westerberg@linux.intel.com> Tested-by:
Wolfram Sang <wsa@the-dreams.de> Cc: stable@kernel.org
-
- 17 Oct, 2015 1 commit
-
-
Linus Torvalds authored
Merge branches 'irq-urgent-for-linus' and 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip Pull irq/timer fixes from Thomas Gleixner: "irq: a fix for the new hierarchical MSI interrupt handling which unbreaks PCI=n configurations. timers: a fix for the new hrtimer clock offset update mechanism to ensure that the boot time offset is respected" * 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: genirq/msi: Do not use pci_msi_[un]mask_irq as default methods * 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: timekeeping: Increment clock_was_set_seq in timekeeping_init()
-
