From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To: linux-kernel@vger.kernel.org
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
stable@vger.kernel.org,
Dennis Dalessandro <dennis.dalessandro@intel.com>,
Mike Marciniszyn <mike.marciniszyn@intel.com>,
Doug Ledford <dledford@redhat.com>
Subject: [PATCH 4.10 48/93] IB/hfi1: Prevent kernel QP post send hard lockups
Date: Thu, 18 May 2017 12:47:17 +0200 [thread overview]
Message-ID: <20170518104745.086140798@linuxfoundation.org> (raw)
In-Reply-To: <20170518104743.163522815@linuxfoundation.org>
4.10-stable review patch. If anyone has any objections, please let me know.
------------------
From: Mike Marciniszyn <mike.marciniszyn@intel.com>
commit b6eac931b9bb2bce4db7032c35b41e5e34ec22a5 upstream.
The driver progress routines can call cond_resched() when
a timeslice is exhausted and irqs are enabled.
If the ULP had been holding a spin lock without disabling irqs and
the post send directly called the progress routine, the cond_resched()
could yield allowing another thread from the same ULP to deadlock
on that same lock.
Correct by replacing the current hfi1_do_send() calldown with a unique
one for post send and adding an argument to hfi1_do_send() to indicate
that the send engine is running in a thread. If the routine is not
running in a thread, avoid calling cond_resched().
Fixes: Commit 831464ce4b74 ("IB/hfi1: Don't call cond_resched in atomic mode when sending packets")
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/infiniband/hw/hfi1/ruc.c | 26 ++++++++++++++++----------
drivers/infiniband/hw/hfi1/verbs.c | 4 ++--
drivers/infiniband/hw/hfi1/verbs.h | 6 ++++--
3 files changed, 22 insertions(+), 14 deletions(-)
--- a/drivers/infiniband/hw/hfi1/ruc.c
+++ b/drivers/infiniband/hw/hfi1/ruc.c
@@ -1,5 +1,5 @@
/*
- * Copyright(c) 2015, 2016 Intel Corporation.
+ * Copyright(c) 2015 - 2017 Intel Corporation.
*
* This file is provided under a dual BSD/GPLv2 license. When using or
* redistributing this file, you may do so under either license.
@@ -823,23 +823,29 @@ void hfi1_make_ruc_header(struct rvt_qp
/* when sending, force a reschedule every one of these periods */
#define SEND_RESCHED_TIMEOUT (5 * HZ) /* 5s in jiffies */
+void hfi1_do_send_from_rvt(struct rvt_qp *qp)
+{
+ hfi1_do_send(qp, false);
+}
+
void _hfi1_do_send(struct work_struct *work)
{
struct iowait *wait = container_of(work, struct iowait, iowork);
struct rvt_qp *qp = iowait_to_qp(wait);
- hfi1_do_send(qp);
+ hfi1_do_send(qp, true);
}
/**
* hfi1_do_send - perform a send on a QP
* @work: contains a pointer to the QP
+ * @in_thread: true if in a workqueue thread
*
* Process entries in the send work queue until credit or queue is
* exhausted. Only allow one CPU to send a packet per QP.
* Otherwise, two threads could send packets out of order.
*/
-void hfi1_do_send(struct rvt_qp *qp)
+void hfi1_do_send(struct rvt_qp *qp, bool in_thread)
{
struct hfi1_pkt_state ps;
struct hfi1_qp_priv *priv = qp->priv;
@@ -907,8 +913,10 @@ void hfi1_do_send(struct rvt_qp *qp)
qp->s_hdrwords = 0;
/* allow other tasks to run */
if (unlikely(time_after(jiffies, timeout))) {
- if (workqueue_congested(cpu,
- ps.ppd->hfi1_wq)) {
+ if (!in_thread ||
+ workqueue_congested(
+ cpu,
+ ps.ppd->hfi1_wq)) {
spin_lock_irqsave(
&qp->s_lock,
ps.flags);
@@ -921,11 +929,9 @@ void hfi1_do_send(struct rvt_qp *qp)
*ps.ppd->dd->send_schedule);
return;
}
- if (!irqs_disabled()) {
- cond_resched();
- this_cpu_inc(
- *ps.ppd->dd->send_schedule);
- }
+ cond_resched();
+ this_cpu_inc(
+ *ps.ppd->dd->send_schedule);
timeout = jiffies + (timeout_int) / 8;
}
spin_lock_irqsave(&qp->s_lock, ps.flags);
--- a/drivers/infiniband/hw/hfi1/verbs.c
+++ b/drivers/infiniband/hw/hfi1/verbs.c
@@ -1,5 +1,5 @@
/*
- * Copyright(c) 2015, 2016 Intel Corporation.
+ * Copyright(c) 2015 - 2017 Intel Corporation.
*
* This file is provided under a dual BSD/GPLv2 license. When using or
* redistributing this file, you may do so under either license.
@@ -1832,7 +1832,7 @@ int hfi1_register_ib_device(struct hfi1_
dd->verbs_dev.rdi.driver_f.qp_priv_free = qp_priv_free;
dd->verbs_dev.rdi.driver_f.free_all_qps = free_all_qps;
dd->verbs_dev.rdi.driver_f.notify_qp_reset = notify_qp_reset;
- dd->verbs_dev.rdi.driver_f.do_send = hfi1_do_send;
+ dd->verbs_dev.rdi.driver_f.do_send = hfi1_do_send_from_rvt;
dd->verbs_dev.rdi.driver_f.schedule_send = hfi1_schedule_send;
dd->verbs_dev.rdi.driver_f.schedule_send_no_lock = _hfi1_schedule_send;
dd->verbs_dev.rdi.driver_f.get_pmtu_from_attr = get_pmtu_from_attr;
--- a/drivers/infiniband/hw/hfi1/verbs.h
+++ b/drivers/infiniband/hw/hfi1/verbs.h
@@ -1,5 +1,5 @@
/*
- * Copyright(c) 2015, 2016 Intel Corporation.
+ * Copyright(c) 2015 - 2017 Intel Corporation.
*
* This file is provided under a dual BSD/GPLv2 license. When using or
* redistributing this file, you may do so under either license.
@@ -370,7 +370,9 @@ void hfi1_make_ruc_header(struct rvt_qp
void _hfi1_do_send(struct work_struct *work);
-void hfi1_do_send(struct rvt_qp *qp);
+void hfi1_do_send_from_rvt(struct rvt_qp *qp);
+
+void hfi1_do_send(struct rvt_qp *qp, bool in_thread);
void hfi1_send_complete(struct rvt_qp *qp, struct rvt_swqe *wqe,
enum ib_wc_status status);
next prev parent reply other threads:[~2017-05-18 10:47 UTC|newest]
Thread overview: 92+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-05-18 10:46 [PATCH 4.10 00/93] 4.10.17-stable review Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 01/93] xen: adjust early dom0 p2m handling to xen hypervisor behavior Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 02/93] target: Fix compare_and_write_callback handling for non GOOD status Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 03/93] target/fileio: Fix zero-length READ and WRITE handling Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 04/93] iscsi-target: Set session_fall_back_to_erl0 when forcing reinstatement Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 05/93] usb: xhci: bInterval quirk for TI TUSB73x0 Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 06/93] usb: host: xhci: print correct command ring address Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 07/93] USB: serial: ftdi_sio: add device ID for Microsemi/Arrow SF2PLUS Dev Kit Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 08/93] USB: Proper handling of Race Condition when two USB class drivers try to call init_usb_class simultaneously Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 10/93] staging: vt6656: use off stack for in buffer USB transfers Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 11/93] staging: vt6656: use off stack for out " Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 12/93] staging: gdm724x: gdm_mux: fix use-after-free on module unload Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 13/93] staging: wilc1000: Fix problem with wrong vif index Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 14/93] staging: comedi: jr3_pci: fix possible null pointer dereference Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 15/93] staging: comedi: jr3_pci: cope with jiffies wraparound Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 16/93] usb: misc: add missing continue in switch Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 17/93] usb: gadget: legacy gadgets are optional Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 18/93] usb: Make sure usb/phy/of gets built-in Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 19/93] usb: hub: Fix error loop seen after hub communication errors Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 20/93] usb: hub: Do not attempt to autosuspend disconnected devices Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 21/93] usb: misc: legousbtower: Fix buffers on stack Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 22/93] x86/boot: Fix BSS corruption/overwrite bug in early x86 kernel startup Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 23/93] selftests/x86/ldt_gdt_32: Work around a glibc sigaction() bug Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 24/93] x86, pmem: Fix cache flushing for iovec write < 8 bytes Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 25/93] um: Fix PTRACE_POKEUSER on x86_64 Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 26/93] perf/x86: Fix Broadwell-EP DRAM RAPL events Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 28/93] KVM: arm/arm64: fix races in kvm_psci_vcpu_on Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 29/93] arm64: KVM: Fix decoding of Rt/Rt2 when trapping AArch32 CP accesses Greg Kroah-Hartman
2017-05-18 10:46 ` [PATCH 4.10 30/93] block: fix blk_integrity_register to use templates interval_exp if not 0 Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 31/93] crypto: s5p-sss - Close possible race for completed requests Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 32/93] crypto: algif_aead - Require setkey before accept(2) Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 33/93] crypto: ccp - Use only the relevant interrupt bits Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 34/93] crypto: ccp - Disable interrupts early on unload Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 35/93] crypto: ccp - Change ISR handler method for a v3 CCP Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 36/93] crypto: ccp - Change ISR handler method for a v5 CCP Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 37/93] dm crypt: rewrite (wipe) key in crypto layer using random data Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 38/93] dm era: save spacemap metadata root after the pre-commit Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 39/93] dm rq: check blk_mq_register_dev() return value in dm_mq_init_request_queue() Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 40/93] dm thin: fix a memory leak when passing discard bio down Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 41/93] vfio/type1: Remove locked page accounting workqueue Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 42/93] iov_iter: dont revert iov buffer if csum error Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 43/93] IB/core: Fix sysfs registration error flow Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 44/93] IB/core: For multicast functions, verify that LIDs are multicast LIDs Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 45/93] IB/IPoIB: ibX: failed to create mcg debug file Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 46/93] IB/mlx4: Fix ib device initialization error flow Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 47/93] IB/mlx4: Reduce SRIOV multicast cleanup warning message to debug level Greg Kroah-Hartman
2017-05-18 10:47 ` Greg Kroah-Hartman [this message]
2017-05-18 10:47 ` [PATCH 4.10 49/93] perf auxtrace: Fix no_size logic in addr_filter__resolve_kernel_syms() Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 50/93] perf annotate s390: Fix perf annotate error -95 (4.10 regression) Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 51/93] perf annotate s390: Implement jump types for perf annotate Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 52/93] jbd2: fix dbench4 performance regression for nobarrier mounts Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 53/93] ext4: evict inline data when writing to memory map Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 54/93] orangefs: fix bounds check for listxattr Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 55/93] orangefs: clean up oversize xattr validation Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 56/93] orangefs: do not set getattr_time on orangefs_lookup Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 57/93] orangefs: do not check possibly stale size on truncate Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 58/93] fs/xattr.c: zero out memory copied to userspace in getxattr Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 59/93] ceph: fix memory leak in __ceph_setxattr() Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 60/93] fs/block_dev: always invalidate cleancache in invalidate_bdev() Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 61/93] mm: prevent potential recursive reclaim due to clearing PF_MEMALLOC Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 62/93] Fix match_prepath() Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 63/93] Set unicode flag on cifs echo request to avoid Mac error Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 64/93] SMB3: Work around mount failure when using SMB3 dialect to Macs Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 66/93] cifs: fix leak in FSCTL_ENUM_SNAPS response handling Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 67/93] cifs: fix CIFS_ENUMERATE_SNAPSHOTS oops Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 68/93] CIFS: fix oplock break deadlocks Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 69/93] cifs: fix CIFS_IOC_GET_MNT_INFO oops Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 71/93] ovl: do not set overlay.opaque on non-dir create Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 72/93] padata: free correct variable Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 73/93] md/raid1: avoid reusing a resync bio after error handling Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 74/93] device-dax: fix cdev leak Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 75/93] device-dax: fix sysfs attribute deadlock Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 76/93] dax: prevent invalidation of mapped DAX entries Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 77/93] mm: fix data corruption due to stale mmap reads Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 78/93] f2fs: fix fs corruption due to zero inode page Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 79/93] fscrypt: fix context consistency check when key(s) unavailable Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 80/93] serial: samsung: Use right device for DMA-mapping calls Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 81/93] serial: omap: fix runtime-pm handling on unbind Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 82/93] serial: omap: suspend device on probe errors Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 83/93] tty: pty: Fix ldisc flush after userspace become aware of the data already Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 84/93] Bluetooth: Fix user channel for 32bit userspace on 64bit kernel Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 85/93] Bluetooth: hci_bcm: add missing tty-device sanity check Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 86/93] Bluetooth: hci_intel: " Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 87/93] ipmi: Fix kernel panic at ipmi_ssif_thread() Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 88/93] libnvdimm, region: fix flush hint detection crash Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 89/93] libnvdimm, pmem: fix a NULL pointer BUG in nd_pmem_notify Greg Kroah-Hartman
2017-05-18 10:47 ` [PATCH 4.10 90/93] libnvdimm: fix nvdimm_bus_lock() vs device_lock() ordering Greg Kroah-Hartman
2017-05-18 10:48 ` [PATCH 4.10 91/93] libnvdimm, pfn: fix npfns vs section alignment Greg Kroah-Hartman
2017-05-18 10:48 ` [PATCH 4.10 92/93] pstore: Fix flags to enable dumps on powerpc Greg Kroah-Hartman
2017-05-18 10:48 ` [PATCH 4.10 93/93] pstore: Shut down worker when unregistering Greg Kroah-Hartman
2017-05-18 17:31 ` [PATCH 4.10 00/93] 4.10.17-stable review Shuah Khan
2017-05-19 1:10 ` Guenter Roeck
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20170518104745.086140798@linuxfoundation.org \
--to=gregkh@linuxfoundation.org \
--cc=dennis.dalessandro@intel.com \
--cc=dledford@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mike.marciniszyn@intel.com \
--cc=stable@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).