If we have a read layout, then sanity check the minimal layout length
so that it does not extend beyond the end of file.
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
According to RFC5661 section 18.43.3, if the server cannot satisfy
the loga_minlength argument to LAYOUTGET, there are 2 cases:
1) If loga_minlength == 0, it returns NFS4ERR_LAYOUTTRYLATER
2) If loga_minlength != 0, it returns NFS4ERR_BADLAYOUT
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
User visible:
- Add new compaction-times python script (Tony Jones)
- Make the --[no-]-demangle/--[no-]-demangle-kernel command line
options available in 'perf script' too (Mark Drayton)
- Allow for negative numbers in libtraceevent's print format,
fixing up misformatting in some tracepoints (Steven Rostedt)
Infrastructure:
- perf_env/perf_evlist changes to allow accessing the data
structure with the environment where some perf data was
collected in functions not necessarily related to perf.data
file processing (Kan Liang)
- Cleanups for the tracepoint definition location paths routines (Jiri Olsa)
- Introduce sysfs/filename__sprintf_build_id, removing code
duplication (Masami Hiramatsu)
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1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=9AFe
-----END PGP SIGNATURE-----
Merge tag 'perf-core-for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/core
Pull perf/core improvement and fixes from Arnaldo Carvalho de Melo:
User visible changes:
- Add new compaction-times python script. (Tony Jones)
- Make the --[no-]-demangle/--[no-]-demangle-kernel command line
options available in 'perf script' too. (Mark Drayton)
- Allow for negative numbers in libtraceevent's print format,
fixing up misformatting in some tracepoints. (Steven Rostedt)
Infrastructure changes:
- perf_env/perf_evlist changes to allow accessing the data
structure with the environment where some perf data was
collected in functions not necessarily related to perf.data
file processing. (Kan Liang)
- Cleanups for the tracepoint definition location paths routines. (Jiri Olsa)
- Introduce sysfs/filename__sprintf_build_id, removing code
duplication. (Masami Hiramatsu)
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
User visible:
- Use index, not CPU id, to find core/pkg id in 'perf stat' (Kan Liang)
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1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=u4R/
-----END PGP SIGNATURE-----
Merge tag 'perf-urgent-for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/urgent
Pull perf/urgent fix from Arnaldo Carvalho de Melo:
- Use index, not CPU id, to find core/pkg id in 'perf stat' (Kan Liang)
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
The following panic is captured in ker3.14, but the issue still exists
in latest kernel.
---------------------------------------------------------------------
[ 20.738217] c0 3136 (Compiler) Unable to handle kernel NULL pointer dereference
at virtual address 00000578
......
[ 20.738499] c0 3136 (Compiler) PC is at _raw_spin_lock_irqsave+0x24/0x60
[ 20.738527] c0 3136 (Compiler) LR is at _raw_spin_lock_irqsave+0x20/0x60
[ 20.740134] c0 3136 (Compiler) Call trace:
[ 20.740165] c0 3136 (Compiler) [<ffffffc0008ee900>] _raw_spin_lock_irqsave+0x24/0x60
[ 20.740200] c0 3136 (Compiler) [<ffffffc0000dd024>] __wake_up+0x1c/0x54
[ 20.740230] c0 3136 (Compiler) [<ffffffc000639414>] mmc_wait_data_done+0x28/0x34
[ 20.740262] c0 3136 (Compiler) [<ffffffc0006391a0>] mmc_request_done+0xa4/0x220
[ 20.740314] c0 3136 (Compiler) [<ffffffc000656894>] sdhci_tasklet_finish+0xac/0x264
[ 20.740352] c0 3136 (Compiler) [<ffffffc0000a2b58>] tasklet_action+0xa0/0x158
[ 20.740382] c0 3136 (Compiler) [<ffffffc0000a2078>] __do_softirq+0x10c/0x2e4
[ 20.740411] c0 3136 (Compiler) [<ffffffc0000a24bc>] irq_exit+0x8c/0xc0
[ 20.740439] c0 3136 (Compiler) [<ffffffc00008489c>] handle_IRQ+0x48/0xac
[ 20.740469] c0 3136 (Compiler) [<ffffffc000081428>] gic_handle_irq+0x38/0x7c
----------------------------------------------------------------------
Because in SMP, "mrq" has race condition between below two paths:
path1: CPU0: <tasklet context>
static void mmc_wait_data_done(struct mmc_request *mrq)
{
mrq->host->context_info.is_done_rcv = true;
//
// If CPU0 has just finished "is_done_rcv = true" in path1, and at
// this moment, IRQ or ICache line missing happens in CPU0.
// What happens in CPU1 (path2)?
//
// If the mmcqd thread in CPU1(path2) hasn't entered to sleep mode:
// path2 would have chance to break from wait_event_interruptible
// in mmc_wait_for_data_req_done and continue to run for next
// mmc_request (mmc_blk_rw_rq_prep).
//
// Within mmc_blk_rq_prep, mrq is cleared to 0.
// If below line still gets host from "mrq" as the result of
// compiler, the panic happens as we traced.
wake_up_interruptible(&mrq->host->context_info.wait);
}
path2: CPU1: <The mmcqd thread runs mmc_queue_thread>
static int mmc_wait_for_data_req_done(...
{
...
while (1) {
wait_event_interruptible(context_info->wait,
(context_info->is_done_rcv ||
context_info->is_new_req));
static void mmc_blk_rw_rq_prep(...
{
...
memset(brq, 0, sizeof(struct mmc_blk_request));
This issue happens very coincidentally; however adding mdelay(1) in
mmc_wait_data_done as below could duplicate it easily.
static void mmc_wait_data_done(struct mmc_request *mrq)
{
mrq->host->context_info.is_done_rcv = true;
+ mdelay(1);
wake_up_interruptible(&mrq->host->context_info.wait);
}
At runtime, IRQ or ICache line missing may just happen at the same place
of the mdelay(1).
This patch gets the mmc_context_info at the beginning of function, it can
avoid this race condition.
Signed-off-by: Jialing Fu <jlfu@marvell.com>
Tested-by: Shawn Lin <shawn.lin@rock-chips.com>
Fixes: 2220eedfd7 ("mmc: fix async request mechanism ....")
Signed-off-by: Shawn Lin <shawn.lin@rock-chips.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
net/ipv4/af_inet.c: In function 'snmp_get_cpu_field64':
>> net/ipv4/af_inet.c:1486:26: error: 'offt' undeclared (first use in this function)
v = *(((u64 *)bhptr) + offt);
^
net/ipv4/af_inet.c:1486:26: note: each undeclared identifier is reported only once for each function it appears in
net/ipv4/af_inet.c: In function 'snmp_fold_field64':
>> net/ipv4/af_inet.c:1499:39: error: 'offct' undeclared (first use in this function)
res += snmp_get_cpu_field(mib, cpu, offct, syncp_offset);
^
>> net/ipv4/af_inet.c:1499:10: error: too many arguments to function 'snmp_get_cpu_field'
res += snmp_get_cpu_field(mib, cpu, offct, syncp_offset);
^
net/ipv4/af_inet.c:1455:5: note: declared here
u64 snmp_get_cpu_field(void __percpu *mib, int cpu, int offt)
^
Reported-by: kbuild test robot <fengguang.wu@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Poll() returns immediately after setting the kernel current frame
(ring->head) to SKIP from user space even though there is no new
frame. And in a case of all frames is VALID, user space program
unintensionally sets (only) kernel current frame to UNUSED, then
calls poll(), it will not return immediately even though there are
VALID frames.
To avoid situations like above, I think we need to scan all frames
to find VALID frames at poll() like netlink_alloc_skb(),
netlink_forward_ring() finding an UNUSED frame at skb allocation.
Signed-off-by: Ken-ichirou MATSUZAWA <chamas@h4.dion.ne.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>
Aleksey Makarov says:
====================
net: thunderx: New features and fixes
v2:
- The unused affinity_mask field of the structure cmp_queue
has been deleted. (thanks to David Miller)
- The unneeded initializers have been dropped. (thanks to Alexey Klimov)
- The commit message "net: thunderx: Rework interrupt handling"
has been fixed. (thanks to Alexey Klimov)
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Support for setting VF's corresponding BGX LMAC in internal
loopback mode. This mode can be used for verifying basic HW
functionality such as packet I/O, RX checksum validation,
CQ/RBDR interrupts, stats e.t.c. Useful when DUT has no external
network connectivity.
'loopback' mode can be enabled or disabled via ethtool.
Note: This feature is not supported when no of VFs enabled are
morethan no of physical interfaces i.e active BGX LMACs
Signed-off-by: Sunil Goutham <sgoutham@cavium.com>
Signed-off-by: Aleksey Makarov <aleksey.makarov@caviumnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds support for handling multiple qsets assigned to a
single VF. There by increasing no of queues from earlier 8 to max
no of CPUs in the system i.e 48 queues on a single node and 96 on
dual node system. User doesn't have option to assign which Qsets/VFs
to be merged. Upon request from VF, PF assigns next free Qsets as
secondary qsets. To maintain current behavior no of queues is kept
to 8 by default which can be increased via ethtool.
If user wants to unbind NICVF driver from a secondary Qset then it
should be done after tearing down primary VF's interface.
Signed-off-by: Sunil Goutham <sgoutham@cavium.com>
Signed-off-by: Aleksey Makarov <aleksey.makarov@caviumnetworks.com>
Signed-off-by: Robert Richter <rrichter@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Rework interrupt handler to avoid checking IRQ affinity of
CQ interrupts. Now separate handlers are registered for each IRQ
including RBDR. Register interrupt handlers for only those
which are being used. Add nicvf_dump_intr_status() and use it
in irq handlers.
Signed-off-by: Sunil Goutham <sgoutham@cavium.com>
Signed-off-by: Aleksey Makarov <aleksey.makarov@caviumnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch configures HW to strip 802.1Q header if found in a
receiving packet. The stripped VLAN ID and TCI information is
passed on to software via CQE_RX. Also sets netdev's 'vlan_features'
so that other HW offload features can be used for tagged packets.
This offload feature can be enabled or disabled via ethtool.
Network stack normally ignores RPS for 802.1Q packets and hence low
throughput. With this offload enabled throughput for tagged packets
will be almost same as normal packets.
Note: This patch doesn't enable HW VLAN insertion for transmit packets.
Signed-off-by: Sunil Goutham <sgoutham@cavium.com>
Signed-off-by: Aleksey Makarov <aleksey.makarov@caviumnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Adding support for receive hashing HW offload by using RSS_ALG
and RSS_TAG fields of CQE_RX descriptor. Also removed dependency
on minimum receive queue count to configure RSS so that hash is
always generated.
This hash is used by RPS logic to distribute flows across multiple
CPUs. Offload can be disabled via ethtool.
Signed-off-by: Robert Richter <rrichter@cavium.com>
Signed-off-by: Sunil Goutham <sgoutham@cavium.com>
Signed-off-by: Aleksey Makarov <aleksey.makarov@caviumnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use the nicvf_send_msg_to_pf() function in the mailbox code.
Signed-off-by: Sunil Goutham <sgoutham@cavium.com>
Signed-off-by: Robert Richter <rrichter@cavium.com>
Signed-off-by: Aleksey Makarov <aleksey.makarov@caviumnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Added ethtool support to dump receive packet error statistics reported
in CQE. Also made some small fixes
Signed-off-by: Sunil Goutham <sgoutham@cavium.com>
Signed-off-by: Aleksey Makarov <aleksey.makarov@caviumnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The liquidio and thunder drivers have different maintainers.
Signed-off-by: Aleksey Makarov <aleksey.makarov@caviumnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Raghavendra K T says:
====================
Optimize the snmp stat aggregation for large cpus
While creating 1000 containers, perf is showing lot of time spent in
snmp_fold_field on a large cpu system.
The current patch tries to improve by reordering the statistics gathering.
Please note that similar overhead was also reported while creating
veth pairs https://lkml.org/lkml/2013/3/19/556
Changes in V4:
- remove 'item' variable and use IPSTATS_MIB_MAX to avoid sparse
warning (Eric) also remove 'item' parameter (Joe)
- add missing memset of padding.
Changes in V3:
- use memset to initialize temp buffer in leaf function. (David)
- use memcpy to copy the buffer data to stat instead of unalign_pu (Joe)
- Move buffer definition to leaf function __snmp6_fill_stats64() (Eric)
-
Changes in V2:
- Allocate the stat calculation buffer in stack. (Eric)
Setup:
160 cpu (20 core) baremetal powerpc system with 1TB memory
1000 docker containers was created with command
docker run -itd ubuntu:15.04 /bin/bash in loop
observation:
Docker container creation linearly increased from around 1.6 sec to 7.5 sec
(at 1000 containers) perf data showed, creating veth interfaces resulting in
the below code path was taking more time.
rtnl_fill_ifinfo
-> inet6_fill_link_af
-> inet6_fill_ifla6_attrs
-> snmp_fold_field
proposed idea:
currently __snmp6_fill_stats64 calls snmp_fold_field that walks
through per cpu data to of an item (iteratively for around 36 items).
The patch tries to aggregate the statistics by going through
all the items of each cpu sequentially which is reducing cache
misses.
Performance of docker creation improved by around more than 2x
after the patch.
before the patch:
================
3f45ba571a42e925c4ec4aaee0e48d7610a9ed82a4c931f83324d41822cf6617
real 0m6.836s
user 0m0.095s
sys 0m0.011s
perf record -a docker run -itd ubuntu:15.04 /bin/bash
=======================================================
50.73% docker [kernel.kallsyms] [k] snmp_fold_field
9.07% swapper [kernel.kallsyms] [k] snooze_loop
3.49% docker [kernel.kallsyms] [k] veth_stats_one
2.85% swapper [kernel.kallsyms] [k] _raw_spin_lock
1.37% docker docker [.] backtrace_qsort
1.31% docker docker [.] strings.FieldsFunc
cache-misses: 2.7%
after the patch:
=============
9178273e9df399c8290b6c196e4aef9273be2876225f63b14a60cf97eacfafb5
real 0m3.249s
user 0m0.088s
sys 0m0.020s
perf record -a docker run -itd ubuntu:15.04 /bin/bash
=======================================================
10.57% docker docker [.] scanblock
8.37% swapper [kernel.kallsyms] [k] snooze_loop
6.91% docker [kernel.kallsyms] [k] snmp_get_cpu_field
6.67% docker [kernel.kallsyms] [k] veth_stats_one
3.96% docker docker [.] runtime_MSpan_Sweep
2.47% docker docker [.] strings.FieldsFunc
cache-misses: 1.41 %
Please let me know if you have suggestions/comments.
Thanks Eric, Joe and David for the comments.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Docker container creation linearly increased from around 1.6 sec to 7.5 sec
(at 1000 containers) and perf data showed 50% ovehead in snmp_fold_field.
reason: currently __snmp6_fill_stats64 calls snmp_fold_field that walks
through per cpu data of an item (iteratively for around 36 items).
idea: This patch tries to aggregate the statistics by going through
all the items of each cpu sequentially which is reducing cache
misses.
Docker creation got faster by more than 2x after the patch.
Result:
Before After
Docker creation time 6.836s 3.25s
cache miss 2.7% 1.41%
perf before:
50.73% docker [kernel.kallsyms] [k] snmp_fold_field
9.07% swapper [kernel.kallsyms] [k] snooze_loop
3.49% docker [kernel.kallsyms] [k] veth_stats_one
2.85% swapper [kernel.kallsyms] [k] _raw_spin_lock
perf after:
10.57% docker docker [.] scanblock
8.37% swapper [kernel.kallsyms] [k] snooze_loop
6.91% docker [kernel.kallsyms] [k] snmp_get_cpu_field
6.67% docker [kernel.kallsyms] [k] veth_stats_one
changes/ideas suggested:
Using buffer in stack (Eric), Usage of memset (David), Using memcpy in
place of unaligned_put (Joe).
Signed-off-by: Raghavendra K T <raghavendra.kt@linux.vnet.ibm.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
According to RFC5661 Section 18.2.4, CLOSE is supposed to return
the zero stateid. This means that nfs_clear_open_stateid_locked()
cannot assume that the result stateid will always match the 'other'
field of the existing open stateid when trying to determine a race
with a parallel OPEN.
Instead, we look at the argument, and check for matches.
Cc: stable@vger.kernel.org # v4.0+
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
On Alpha we have spinlocks that are 32b in size and an efficient
cmpxchg64 implementation, so we qualify to make use of cmpxchg backed
lockrefs. Select the ARCH_USE_CMPXCHG_LOCKREF Kconfig symbol and provide
a trivial implementation of arch_spin_value_unlocked to satisfy the
lockref code.
Using Linus' simple testcase from
http://article.gmane.org/gmane.linux.file-systems/77466 on a dual CPU
ES47 system I see around an 8% gain:
N Min Max Median Avg Stddev
x 30 6194580 6295654 6272504 6272514 17694.232
+ 30 6731164 6786334 6767982 6764274 13738.863
Difference at 95.0% confidence
491760 +/- 8188.17
7.83992% +/- 0.130541%
(Student's t, pooled s = 15840.5)
Signed-off-by: Matt Turner <mattst88@gmail.com>
This is a second pull-request which adds last part of
atomic modeset/pageflip support, render node support,
clean-up, and fix-up.
* 'exynos-drm-next' of git://git.kernel.org/pub/scm/linux/kernel/git/daeinki/drm-exynos:
drm/exynos: fix build warning to exynos_drm_gem.c
drm/exynos: Properly report supported formats for each device
drm/exynos: add render node support
drm/exynos: implement atomic_{begin/flush} of DECON
drm/exynos: remove legacy ->suspend()/resume()
drm/exynos: Enable atomic modesetting feature
drm/exynos: remove wait queue for pending page flip
drm/exynos: wait all planes updates to finish
drm/exynos: add atomic asynchronous commit
drm/exynos: fimd: only finish update if START == START_S
drm/exynos: add macro to get the address of START_S reg
drm/exynos: check for pending fb before finish update
drm/exynos: fimd: move window protect code to prepare/cleanup_plane
drm/exynos: add prepare and cleanup phases for planes
drm/exynos: fimd: unify call to exynos_drm_crtc_finish_pageflip()
drm/exynos: don't track enabled state at exynos_crtc
Some i915 fixes headed for v4.3. SKL DDI-E is a wip, but here's the
first in a series.
* tag 'drm-intel-next-fixes-2015-08-28' of git://anongit.freedesktop.org/drm-intel:
drm/i915/skl: enable DDI-E hotplug
drm/i915: Fix build warning on 32-bit
drm/i915/skl: Update DDI buffer translation programming.
drm/i915: Allow parsing of variable size child device entries from VBT
drm/i915: fix link rates reported for SKL
drm/i915: fix VBT parsing for SDVO child device mapping
Just one small fix before 4.3 merge window:
- Use linux/mman.h instead of uapi's mman-common.h inside the driver.
* tag 'drm-amdkfd-next-fixes-2015-08-30' of git://people.freedesktop.org/~gabbayo/linux:
amdkfd: use <linux/mman.h> instead of <uapi/asm-generic/mman-common.h>
Currently, IB/cma remove_one flow blocks until all user descriptor managed by
IB/ucma are released. This prevents hot-removal of IB devices. This patch
allows IB/cma to remove devices regardless of user space activity. Upon getting
the RDMA_CM_EVENT_DEVICE_REMOVAL event we close all the underlying HW resources
for the given ucontext. The ucontext itself is still alive till its explicit
destroying by its creator.
Running applications at that time will have some zombie device, further
operations may fail.
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Reviewed-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Implements the IB core disassociate_ucontext API. The driver detaches the HW
resources for a given user context to prevent a dependency between application
termination and device disconnecting. This is done by managing the VMAs that
were mapped to the HW bars such as door bell and blueflame. When need to detach
remap them to an arbitrary kernel page returned by the zap API.
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Jack Morgenstein <jackm@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Enables the uverbs_remove_one to succeed despite the fact that there are
running IB applications working with the given ib device. This
functionality enables a HW device to be unbind/reset despite the fact that
there are running user space applications using it.
It exposes a new IB kernel API named 'disassociate_ucontext' which lets
a driver detaching its HW resources from a given user context without
crashing/terminating the application. In case a driver implemented the
above API and registered with ib_uverb there will be no dependency between its
device to its uverbs_device. Upon calling remove_one of ib_uverbs the call
should return after disassociating the open HW resources without waiting to
clients disconnecting. In case driver didn't implement this API there will be no
change to current behaviour and uverbs_remove_one will return only when last
client has disconnected and reference count on uverbs device became 0.
In case the lower driver device was removed any application will
continue working over some zombie HCA, further calls will ended with an
immediate error.
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Done in preparation for deploying RCU for the device removal
flow. Allows isolating the RCU handling to the uverb_main layer and
keeping the uverbs_cmd code as is.
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Fixes: 2a72f21226 ("IB/uverbs: Remove dev_table")
Before this commit there was a device look-up table that was protected
by a spin_lock used by ib_uverbs_open and by ib_uverbs_remove_one. When
it was dropped and container_of was used instead, it enabled the race
with remove_one as dev might be freed just after:
dev = container_of(inode->i_cdev, struct ib_uverbs_device, cdev) but
before the kref_get.
In addition, this buggy patch added some dead code as
container_of(x,y,z) can never be NULL and so dev can never be NULL.
As a result the comment above ib_uverbs_open saying "the open method
will either immediately run -ENXIO" is wrong as it can never happen.
The solution follows Jason Gunthorpe suggestion from below URL:
https://www.mail-archive.com/linux-rdma@vger.kernel.org/msg25692.html
cdev will hold a kref on the parent (the containing structure,
ib_uverbs_device) and only when that kref is released it is
guaranteed that open will never be called again.
In addition, fixes the active count scheme to use an atomic
not a kref to prevent WARN_ON as pointed by above comment
from Jason.
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Fix the reference counting usage to be handled in the event file
creation/destruction function, instead of being done by the caller.
This is done for both async/non-async event files.
Based on Jason Gunthorpe report at https://www.mail-archive.com/
linux-rdma@vger.kernel.org/msg24680.html:
"The existing code for this is broken, in ib_uverbs_get_context all
the error paths between ib_uverbs_alloc_event_file and the
kref_get(file->ref) are wrong - this will result in fput() which will
call ib_uverbs_event_close, which will try to do kref_put and
ib_unregister_event_handler - which are no longer paired."
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The majority of callers never check the return value, and even if they
did, they can't do anything about a failure.
All possible failure cases represent a bug in the caller, so just
WARN_ON inside the function instead.
This fixes a few random errors:
net/rd/iw.c infinite loops while it fails. (racing with EBUSY?)
This also lays the ground work to get rid of error return from the
drivers. Most drivers do not error, the few that do are broken since
it cannot be handled.
Since uverbs can legitimately make use of EBUSY, open code the
check.
Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Reviewed-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The SRP initiator only needs this if the insecure register_always=N
performance optimization is enabled, or if FRWR/FMR is not supported
in the driver.
Do not create an all physical MR unless it is needed to support
either of those modes. Default register_always to true so the out of
the box configuration does not create an insecure all physical MR.
Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
[bvanassche: reworked and rebased this patch]
Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Instead of always using the global rkey for the indirect data
buffer descriptor, register that descriptor with the HCA if
the kernel module parameter register_always has been set to Y.
Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Introduce the variable srp_device.use_fmr. Leave out the dev->has_fr /
dev->has_fmr and ch->fr_pool / ch->fmr_pool checks since these are
redundant. This patch does not change any functionality but makes the
source code easier to read.
Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Move the srp_map_desc() call from inside srp_map_sg_entry() to
srp_map_sg() such that the use_mr argument can be removed from
srp_map_sg_entry().
Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Mapping a discontiguous sg-list requires multiple memory regions
and hence can exhaust the memory region pool. The SRP initiator
already handles this by temporarily reducing the queue depth. This
means that it is safe to remove the memory registration backtracking
code. This patch has been tested with direct I/O sizes up to 256 MB.
Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Although most paths through which a request is submitted check
block layer parameters like the max_segments limit, these are
not checked when an SG_IO or direct I/O request is submitted.
Hence add a range check for the memory descriptor array pointer.
Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Instead of using the global rkey for large memory regions, use
multiple registrations. See also the while (dma_len) loop further
down in srp_map_sg_entry().
Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
During a discussion in 2011 nobody recalled why FMR was not used for
non-page aligned buffers (see also
http://thread.gmane.org/gmane.linux.drivers.rdma/7149). Re-enable FMR
for such buffers. For the reason why the srp_map_fmr() function needs
to be modified, see also patch "IB/srp: rework mapping engine to use
multiple FMR entries" (commit ID 8f26c9ff9c; January 2011).
Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The pd now has a local_dma_lkey member which completely replaces
ib_get_dma_mr, use it instead.
Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The pd now has a local_dma_lkey member which completely replaces
ib_get_dma_mr, use it instead.
Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Tested-by: Dominique Martinet <dominique.martinet@cea.fr>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The pd now has a local_dma_lkey member which completely replaces
ib_get_dma_mr, use it instead.
Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Reviewed-by: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Replace all leys with pd->local_dma_lkey. This driver does not support
iWarp, so this is safe.
The insecure use of ib_get_dma_mr is thus isolated to an rkey, and will
have to be fixed separately.
Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Reviewed-by: Sagi Grimberg <sagig@mellanox.com>
Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The pd now has a local_dma_lkey member which completely replaces
ib_get_dma_mr, use it instead.
Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Reviewed-by: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Replace all leys with pd->local_dma_lkey. This driver does not support
iWarp, so this is safe.
The insecure use of ib_get_dma_mr is thus isolated to an rkey, and this
looks trivially fixed by forcing the use of registration in a future
patch.
Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Reviewed-by: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The pd now has a local_dma_lkey member which completely replaces
ib_get_dma_mr, use it instead.
Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
