| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
It's quite questionable whether genpd internally should care about if the
corresponding PM domain for a device is powered on, as to allow setting a
new performance state for it. The assumptions creates an unnecessary
limitation at this point, for both consumers and providers, but more
importantly it also makes the code more complicated.
Therefore, let's simplify the code to allow setting a performance state, by
invoking the ->set_performance_state() callback, no matter whether the PM
domain is powered on or off.
Do note, this change means genpd providers needs to restore the performance
state themselves during power on, via the ->power_on() callback. Moreover,
they may also need to check that the PM domain is powered on, from their
->set_performance_state() callback, before deciding to update the state.
Tested-by: Rajendra Nayak <rnayak@codeaurora.org>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
This isn't used anymore, remove it.
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
The OPP core already has the performance state values for each of the
genpd's OPPs and there is no need to call the genpd callback again to
get the performance state for the case where the end device doesn't have
an OPP table and has the "required-opps" property directly in its node.
This commit renames of_genpd_opp_to_performance_state() as
of_get_required_opp_performance_state() and moves it to the OPP core, as
it is all about OPP stuff now.
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
Now that all the infrastructure is in place to support multiple required
OPPs, lets switch over to using it.
A new internal routine _set_required_opps() takes care of updating
performance state for all the required OPPs. With this the performance
state updates are supported even when the end device needs to configure
regulators as well, that wasn't the case earlier.
The pstates were earlier stored in the end device's OPP structures, that
also changes now as those values are stored in the genpd's OPP
structures. And so we switch over to using
pm_genpd_opp_to_performance_state() instead of
of_genpd_opp_to_performance_state() to get performance state for the
genpd OPPs.
The routine _generic_set_opp_domain() is not required anymore and is
removed.
On errors we don't try to recover by reverting to old settings as things
are really complex now and the calls here should never really fail
unless there is a bug. There is no point increasing the complexity, for
code which will never be executed.
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
Multiple generic power domains for a consumer device are supported with
the help of virtual devices, which are created for each consumer device
- genpd pair. These are the device structures which are attached to the
power domain and are required by the OPP core to set the performance
state of the genpd.
The helpers added by this commit are required to be called once for each
of these virtual devices. These are required only if multiple domains
are available for a device, otherwise the actual device structure will
be used instead by the OPP core.
The new helpers also support the complex cases where the consumer device
wouldn't always require all the domains. For example, a camera may
require only one power domain during normal operations but two during
high resolution operations. The consumer driver can call
dev_pm_opp_put_genpd_virt_dev(high_resolution_genpd_virt_dev) if it is
currently operating in the normal mode and doesn't have any performance
requirements from the genpd which manages high resolution power
requirements. The consumer driver can later call
dev_pm_opp_set_genpd_virt_dev(high_resolution_genpd_virt_dev) once it
switches back to the high resolution mode.
The new helpers differ from other OPP set/put helpers as the new ones
can be called with OPPs initialized for the table as we may need to call
them on the fly because of the complex case explained above. For this
reason it is possible that the genpd virt_dev structure may be used in
parallel while the new helpers are running and a new mutex is added to
protect against that. We didn't use the existing opp_table->lock mutex
as that is widely used in the OPP core and we will need this lock in the
dev_pm_opp_set_rate() helper while changing OPP and we need to make sure
there is not much contention while doing that as that's the hotpath.
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
The OPP core currently stores the performance state in the consumer
device's OPP table, but that is going to change going forward and
performance state will rather be set directly in the genpd's OPP table.
For that we need to get the performance state for genpd's device
structure (genpd->dev) instead of the consumer device's structure. Add a
new helper to do that.
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
An earlier commit populated the OPP tables from the "required-opps"
property, this commit populates the individual OPPs. This is repeated
for each OPP in the OPP table and these populated OPPs will be used by
later commits.
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
The current implementation works only for the case where a single
phandle is present in the "required-opps" property, while DT allows
multiple phandles to be present there.
This patch adds new infrastructure to parse all the phandles present in
"required-opps" property and save pointers of the required OPP's OPP
tables. These will be used by later commits.
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
Create a separate routine to take care of custom set_opp() handler
specific stuff.
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
We need to handle genpd OPP tables differently, this is already the case
at one location and will be extended going forward. Add another field to
the OPP table to check if the table belongs to a genpd or not.
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
There are several struct device instances that genpd core handles. The
most common one is the consumer device structure, which is named
(correctly) as "dev" within genpd core. The second one is the genpd's
device structure, referenced as genpd->dev. The third one is the virtual
device structures created by the genpd core to represent the consumer
device for multiple power domain case, currently named as genpd_dev. The
naming of these virtual devices isn't very clear or readable and it
looks more like the genpd->dev.
Rename the virtual device instances within the genpd core as "virt_dev".
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
|
|
|
|
With the addition of the NUMA identity level, we increased @level by
one and will run off the end of the array in the distance sort loop.
Fixed: 051f3ca02e46 ("sched/topology: Introduce NUMA identity node sched domain")
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
|
|
We have received a bug report that unbinding a large pmem (>1TB) can
result in a soft lockup:
NMI watchdog: BUG: soft lockup - CPU#9 stuck for 23s! [ndctl:4365]
[...]
Supported: Yes
CPU: 9 PID: 4365 Comm: ndctl Not tainted 4.12.14-94.40-default #1 SLE12-SP4
Hardware name: Intel Corporation S2600WFD/S2600WFD, BIOS SE5C620.86B.01.00.0833.051120182255 05/11/2018
task: ffff9cce7d4410c0 task.stack: ffffbe9eb1bc4000
RIP: 0010:__put_page+0x62/0x80
Call Trace:
devm_memremap_pages_release+0x152/0x260
release_nodes+0x18d/0x1d0
device_release_driver_internal+0x160/0x210
unbind_store+0xb3/0xe0
kernfs_fop_write+0x102/0x180
__vfs_write+0x26/0x150
vfs_write+0xad/0x1a0
SyS_write+0x42/0x90
do_syscall_64+0x74/0x150
entry_SYSCALL_64_after_hwframe+0x3d/0xa2
RIP: 0033:0x7fd13166b3d0
It has been reported on an older (4.12) kernel but the current upstream
code doesn't cond_resched in the hot remove code at all and the given
range to remove might be really large. Fix the issue by calling
cond_resched once per memory section.
Link: http://lkml.kernel.org/r/20181031125840.23982-1-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Johannes Thumshirn <jthumshirn@suse.de>
Cc: Dan Williams <dan.j.williams@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
syzbot is reporting too large memory allocation at bfs_fill_super() [1].
Since file system image is corrupted such that bfs_sb->s_start == 0,
bfs_fill_super() is trying to allocate 8MB of continuous memory. Fix
this by adding a sanity check on bfs_sb->s_start, __GFP_NOWARN and
printf().
[1] https://syzkaller.appspot.com/bug?id=16a87c236b951351374a84c8a32f40edbc034e96
Link: http://lkml.kernel.org/r/1525862104-3407-1-git-send-email-penguin-kernel@I-love.SAKURA.ne.jp
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Reported-by: syzbot <syzbot+71c6b5d68e91149fc8a4@syzkaller.appspotmail.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Tigran Aivazian <aivazian.tigran@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
Remove one include of <linux/pipe_fs_i.h>.
No functional changes.
Link: http://lkml.kernel.org/r/20181004134223.17735-1-michael@schupikov.de
Signed-off-by: Michael Schupikov <michael@schupikov.de>
Reviewed-by: Richard Weinberger <richard@nod.at>
Acked-by: Luis Chamberlain <mcgrof@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
We include kexec.h and slab.h twice in kexec_file.c. It's unnecessary.
hence just remove them.
Link: http://lkml.kernel.org/r/1537498098-19171-1-git-send-email-zhongjiang@huawei.com
Signed-off-by: zhong jiang <zhongjiang@huawei.com>
Reviewed-by: Bhupesh Sharma <bhsharma@redhat.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Baoquan He <bhe@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
THP allocation mode is quite complex and it depends on the defrag mode.
This complexity is hidden in alloc_hugepage_direct_gfpmask from a large
part currently. The NUMA special casing (namely __GFP_THISNODE) is
however independent and placed in alloc_pages_vma currently. This both
adds an unnecessary branch to all vma based page allocation requests and
it makes the code more complex unnecessarily as well. Not to mention
that e.g. shmem THP used to do the node reclaiming unconditionally
regardless of the defrag mode until recently. This was not only
unexpected behavior but it was also hardly a good default behavior and I
strongly suspect it was just a side effect of the code sharing more than
a deliberate decision which suggests that such a layering is wrong.
Get rid of the thp special casing from alloc_pages_vma and move the
logic to alloc_hugepage_direct_gfpmask. __GFP_THISNODE is applied to the
resulting gfp mask only when the direct reclaim is not requested and
when there is no explicit numa binding to preserve the current logic.
Please note that there's also a slight difference wrt MPOL_BIND now. The
previous code would avoid using __GFP_THISNODE if the local node was
outside of policy_nodemask(). After this patch __GFP_THISNODE is avoided
for all MPOL_BIND policies. So there's a difference that if local node
is actually allowed by the bind policy's nodemask, previously
__GFP_THISNODE would be added, but now it won't be. From the behavior
POV this is still correct because the policy nodemask is used.
Link: http://lkml.kernel.org/r/20180925120326.24392-3-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Alex Williamson <alex.williamson@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: David Rientjes <rientjes@google.com>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Stefan Priebe - Profihost AG <s.priebe@profihost.ag>
Cc: Zi Yan <zi.yan@cs.rutgers.edu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
ocfs2_defrag_extent() might leak allocated clusters. When the file
system has insufficient space, the number of claimed clusters might be
less than the caller wants. If that happens, the original code might
directly commit the transaction without returning clusters.
This patch is based on code in ocfs2_add_clusters_in_btree().
[akpm@linux-foundation.org: include localalloc.h, reduce scope of data_ac]
Link: http://lkml.kernel.org/r/20180904041621.16874-3-lchen@suse.com
Signed-off-by: Larry Chen <lchen@suse.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Joseph Qi <jiangqi903@gmail.com>
Cc: Changwei Ge <ge.changwei@h3c.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
The handling of timestamps outside of the 1970..2038 range in the dlm
glue is rather inconsistent: on 32-bit architectures, this has always
wrapped around to negative timestamps in the 1902..1969 range, while on
64-bit kernels all timestamps are interpreted as positive 34 bit numbers
in the 1970..2514 year range.
Now that the VFS code handles 64-bit timestamps on all architectures, we
can make the behavior more consistent here, and return the same result
that we had on 64-bit already, making the file system y2038 safe in the
process. Outside of dlmglue, it already uses 64-bit on-disk timestamps
anway, so that part is fine.
For consistency, I'm changing ocfs2_pack_timespec() to clamp anything
outside of the supported range to the minimum and maximum values. This
avoids a possible ambiguity of values before 1970 in particular, which
used to be interpreted as times at the end of the 2514 range previously.
Link: http://lkml.kernel.org/r/20180619155826.4106487-1-arnd@arndb.de
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Joseph Qi <jiangqi903@gmail.com>
Cc: Changwei Ge <ge.changwei@h3c.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
ocfs2_read_blocks() and ocfs2_read_blocks_sync() are both used to read
several blocks from disk. Currently, the input argument *bhs* can be
NULL or NOT. It depends on the caller's behavior. If the function
fails in reading blocks from disk, the corresponding bh will be assigned
to NULL and put.
Obviously, above process for non-NULL input bh is not appropriate.
Because the caller doesn't even know its bhs are put and re-assigned.
If buffer head is managed by caller, ocfs2_read_blocks and
ocfs2_read_blocks_sync() should not evaluate it to NULL. It will cause
caller accessing illegal memory, thus crash.
Link: http://lkml.kernel.org/r/HK2PR06MB045285E0F4FBB561F9F2F9B3D5680@HK2PR06MB0452.apcprd06.prod.outlook.com
Signed-off-by: Changwei Ge <ge.changwei@h3c.com>
Reviewed-by: Guozhonghua <guozhonghua@h3c.com>
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Joseph Qi <jiangqi903@gmail.com>
Cc: Changwei Ge <ge.changwei@h3c.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
Somehow, file system metadata was corrupted, which causes
ocfs2_check_dir_entry() to fail in function ocfs2_dir_foreach_blk_el().
According to the original design intention, if above happens we should
skip the problematic block and continue to retrieve dir entry. But
there is obviouse misuse of brelse around related code.
After failure of ocfs2_check_dir_entry(), current code just moves to
next position and uses the problematic buffer head again and again
during which the problematic buffer head is released for multiple times.
I suppose, this a serious issue which is long-lived in ocfs2. This may
cause other file systems which is also used in a the same host insane.
So we should also consider about bakcporting this patch into linux
-stable.
Link: http://lkml.kernel.org/r/HK2PR06MB045211675B43EED794E597B6D56E0@HK2PR06MB0452.apcprd06.prod.outlook.com
Signed-off-by: Changwei Ge <ge.changwei@h3c.com>
Suggested-by: Changkuo Shi <shi.changkuo@h3c.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Joseph Qi <jiangqi903@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
When -EIOCBQUEUED returns, it means that aio_complete() will be called
from dio_complete(), which is an asynchronous progress against
write_iter. Generally, IO is a very slow progress than executing
instruction, but we still can't take the risk to access a freed iocb.
And we do face a BUG crash issue. Using the crash tool, iocb is
obviously freed already.
crash> struct -x kiocb ffff881a350f5900
struct kiocb {
ki_filp = 0xffff881a350f5a80,
ki_pos = 0x0,
ki_complete = 0x0,
private = 0x0,
ki_flags = 0x0
}
And the backtrace shows:
ocfs2_file_write_iter+0xcaa/0xd00 [ocfs2]
aio_run_iocb+0x229/0x2f0
do_io_submit+0x291/0x540
SyS_io_submit+0x10/0x20
system_call_fastpath+0x16/0x75
Link: http://lkml.kernel.org/r/1523361653-14439-1-git-send-email-ge.changwei@h3c.com
Signed-off-by: Changwei Ge <ge.changwei@h3c.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Joseph Qi <jiangqi903@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
During one dead node's recovery by other node, quota recovery work will
be queued. We should avoid calling quota when it is not supported, so
check the quota flags.
Link: http://lkml.kernel.org/r/71604351584F6A4EBAE558C676F37CA401071AC9FB@H3CMLB12-EX.srv.huawei-3com.com
Signed-off-by: guozhonghua <guozhonghua@h3c.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Joseph Qi <jiangqi903@gmail.com>
Cc: Changwei Ge <ge.changwei@h3c.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
Remove ocfs2_is_o2cb_active(). We have similar functions to identify
which cluster stack is being used via osb->osb_cluster_stack.
Secondly, the current implementation of ocfs2_is_o2cb_active() is not
totally safe. Based on the design of stackglue, we need to get
ocfs2_stack_lock before using ocfs2_stack related data structures, and
that active_stack pointer can be NULL in the case of mount failure.
Link: http://lkml.kernel.org/r/1495441079-11708-1-git-send-email-ghe@suse.com
Signed-off-by: Gang He <ghe@suse.com>
Reviewed-by: Joseph Qi <jiangqi903@gmail.com>
Reviewed-by: Eric Ren <zren@suse.com>
Acked-by: Changwei Ge <ge.changwei@h3c.com>
Cc: Mark Fasheh <mark@fasheh.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
THP allocation might be really disruptive when allocated on NUMA system
with the local node full or hard to reclaim. Stefan has posted an
allocation stall report on 4.12 based SLES kernel which suggests the
same issue:
kvm: page allocation stalls for 194572ms, order:9, mode:0x4740ca(__GFP_HIGHMEM|__GFP_IO|__GFP_FS|__GFP_COMP|__GFP_NOMEMALLOC|__GFP_HARDWALL|__GFP_THISNODE|__GFP_MOVABLE|__GFP_DIRECT_RECLAIM), nodemask=(null)
kvm cpuset=/ mems_allowed=0-1
CPU: 10 PID: 84752 Comm: kvm Tainted: G W 4.12.0+98-ph <a href="/view.php?id=1" title="[geschlossen] Integration Ramdisk" class="resolved">0000001</a> SLE15 (unreleased)
Hardware name: Supermicro SYS-1029P-WTRT/X11DDW-NT, BIOS 2.0 12/05/2017
Call Trace:
dump_stack+0x5c/0x84
warn_alloc+0xe0/0x180
__alloc_pages_slowpath+0x820/0xc90
__alloc_pages_nodemask+0x1cc/0x210
alloc_pages_vma+0x1e5/0x280
do_huge_pmd_wp_page+0x83f/0xf00
__handle_mm_fault+0x93d/0x1060
handle_mm_fault+0xc6/0x1b0
__do_page_fault+0x230/0x430
do_page_fault+0x2a/0x70
page_fault+0x7b/0x80
[...]
Mem-Info:
active_anon:126315487 inactive_anon:1612476 isolated_anon:5
active_file:60183 inactive_file:245285 isolated_file:0
unevictable:15657 dirty:286 writeback:1 unstable:0
slab_reclaimable:75543 slab_unreclaimable:2509111
mapped:81814 shmem:31764 pagetables:370616 bounce:0
free:32294031 free_pcp:6233 free_cma:0
Node 0 active_anon:254680388kB inactive_anon:1112760kB active_file:240648kB inactive_file:981168kB unevictable:13368kB isolated(anon):0kB isolated(file):0kB mapped:280240kB dirty:1144kB writeback:0kB shmem:95832kB shmem_thp: 0kB shmem_pmdmapped: 0kB anon_thp: 81225728kB writeback_tmp:0kB unstable:0kB all_unreclaimable? no
Node 1 active_anon:250583072kB inactive_anon:5337144kB active_file:84kB inactive_file:0kB unevictable:49260kB isolated(anon):20kB isolated(file):0kB mapped:47016kB dirty:0kB writeback:4kB shmem:31224kB shmem_thp: 0kB shmem_pmdmapped: 0kB anon_thp: 31897600kB writeback_tmp:0kB unstable:0kB all_unreclaimable? no
The defrag mode is "madvise" and from the above report it is clear that
the THP has been allocated for MADV_HUGEPAGA vma.
Andrea has identified that the main source of the problem is
__GFP_THISNODE usage:
: The problem is that direct compaction combined with the NUMA
: __GFP_THISNODE logic in mempolicy.c is telling reclaim to swap very
: hard the local node, instead of failing the allocation if there's no
: THP available in the local node.
:
: Such logic was ok until __GFP_THISNODE was added to the THP allocation
: path even with MPOL_DEFAULT.
:
: The idea behind the __GFP_THISNODE addition, is that it is better to
: provide local memory in PAGE_SIZE units than to use remote NUMA THP
: backed memory. That largely depends on the remote latency though, on
: threadrippers for example the overhead is relatively low in my
: experience.
:
: The combination of __GFP_THISNODE and __GFP_DIRECT_RECLAIM results in
: extremely slow qemu startup with vfio, if the VM is larger than the
: size of one host NUMA node. This is because it will try very hard to
: unsuccessfully swapout get_user_pages pinned pages as result of the
: __GFP_THISNODE being set, instead of falling back to PAGE_SIZE
: allocations and instead of trying to allocate THP on other nodes (it
: would be even worse without vfio type1 GUP pins of course, except it'd
: be swapping heavily instead).
Fix this by removing __GFP_THISNODE for THP requests which are
requesting the direct reclaim. This effectivelly reverts 5265047ac301
on the grounds that the zone/node reclaim was known to be disruptive due
to premature reclaim when there was memory free. While it made sense at
the time for HPC workloads without NUMA awareness on rare machines, it
was ultimately harmful in the majority of cases. The existing behaviour
is similar, if not as widespare as it applies to a corner case but
crucially, it cannot be tuned around like zone_reclaim_mode can. The
default behaviour should always be to cause the least harm for the
common case.
If there are specialised use cases out there that want zone_reclaim_mode
in specific cases, then it can be built on top. Longterm we should
consider a memory policy which allows for the node reclaim like behavior
for the specific memory ranges which would allow a
[1] http://lkml.kernel.org/r/20180820032204.9591-1-aarcange@redhat.com
Mel said:
: Both patches look correct to me but I'm responding to this one because
: it's the fix. The change makes sense and moves further away from the
: severe stalling behaviour we used to see with both THP and zone reclaim
: mode.
:
: I put together a basic experiment with usemem configured to reference a
: buffer multiple times that is 80% the size of main memory on a 2-socket
: box with symmetric node sizes and defrag set to "always". The defrag
: setting is not the default but it would be functionally similar to
: accessing a buffer with madvise(MADV_HUGEPAGE). Usemem is configured to
: reference the buffer multiple times and while it's not an interesting
: workload, it would be expected to complete reasonably quickly as it fits
: within memory. The results were;
:
: usemem
: vanilla noreclaim-v1
: Amean Elapsd-1 42.78 ( 0.00%) 26.87 ( 37.18%)
: Amean Elapsd-3 27.55 ( 0.00%) 7.44 ( 73.00%)
: Amean Elapsd-4 5.72 ( 0.00%) 5.69 ( 0.45%)
:
: This shows the elapsed time in seconds for 1 thread, 3 threads and 4
: threads referencing buffers 80% the size of memory. With the patches
: applied, it's 37.18% faster for the single thread and 73% faster with two
: threads. Note that 4 threads showing little difference does not indicate
: the problem is related to thread counts. It's simply the case that 4
: threads gets spread so their workload mostly fits in one node.
:
: The overall view from /proc/vmstats is more startling
:
: 4.19.0-rc1 4.19.0-rc1
: vanillanoreclaim-v1r1
: Minor Faults 35593425 708164
: Major Faults 484088 36
: Swap Ins 3772837 0
: Swap Outs 3932295 0
:
: Massive amounts of swap in/out without the patch
:
: Direct pages scanned 6013214 0
: Kswapd pages scanned 0 0
: Kswapd pages reclaimed 0 0
: Direct pages reclaimed 4033009 0
:
: Lots of reclaim activity without the patch
:
: Kswapd efficiency 100% 100%
: Kswapd velocity 0.000 0.000
: Direct efficiency 67% 100%
: Direct velocity 11191.956 0.000
:
: Mostly from direct reclaim context as you'd expect without the patch.
:
: Page writes by reclaim 3932314.000 0.000
: Page writes file 19 0
: Page writes anon 3932295 0
: Page reclaim immediate 42336 0
:
: Writes from reclaim context is never good but the patch eliminates it.
:
: We should never have default behaviour to thrash the system for such a
: basic workload. If zone reclaim mode behaviour is ever desired but on a
: single task instead of a global basis then the sensible option is to build
: a mempolicy that enforces that behaviour.
This was a severe regression compared to previous kernels that made
important workloads unusable and it starts when __GFP_THISNODE was
added to THP allocations under MADV_HUGEPAGE. It is not a significant
risk to go to the previous behavior before __GFP_THISNODE was added, it
worked like that for years.
This was simply an optimization to some lucky workloads that can fit in
a single node, but it ended up breaking the VM for others that can't
possibly fit in a single node, so going back is safe.
[mhocko@suse.com: rewrote the changelog based on the one from Andrea]
Link: http://lkml.kernel.org/r/20180925120326.24392-2-mhocko@kernel.org
Fixes: 5265047ac301 ("mm, thp: really limit transparent hugepage allocation to local node")
Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Michal Hocko <mhocko@suse.com>
Reported-by: Stefan Priebe <s.priebe@profihost.ag>
Debugged-by: Andrea Arcangeli <aarcange@redhat.com>
Reported-by: Alex Williamson <alex.williamson@redhat.com>
Reviewed-by: Mel Gorman <mgorman@techsingularity.net>
Tested-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Zi Yan <zi.yan@cs.rutgers.edu>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Cc: <stable@vger.kernel.org> [4.1+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
ctags indexing ("make tags" command) throws this warning:
ctags: Warning: include/linux/notifier.h:125:
null expansion of name pattern "\1"
This is the result of DEFINE_PER_CPU() macro expansion. Fix that by
getting rid of line break.
Similar fix was already done in commit 25528213fe9f ("tags: Fix
DEFINE_PER_CPU expansions"), but this one probably wasn't noticed.
Link: http://lkml.kernel.org/r/20181030202808.28027-1-semen.protsenko@linaro.org
Fixes: 9c80172b902d ("kernel/SRCU: provide a static initializer")
Signed-off-by: Sam Protsenko <semen.protsenko@linaro.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
Mike Galbraith reported a regression caused by the commit 9b6f7e163cd0
("mm: rework memcg kernel stack accounting") on a system with
"cgroup_disable=memory" boot option: the system panics with the following
stack trace:
BUG: unable to handle kernel NULL pointer dereference at 00000000000000f8
PGD 0 P4D 0
Oops: 0002 [#1] PREEMPT SMP PTI
CPU: 0 PID: 1 Comm: systemd Not tainted 4.19.0-preempt+ #410
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS ?-20180531_142017-buildhw-08.phx2.fed4
RIP: 0010:page_counter_try_charge+0x22/0xc0
Code: 41 5d c3 c3 0f 1f 40 00 0f 1f 44 00 00 48 85 ff 0f 84 a7 00 00 00 41 56 48 89 f8 49 89 fe 49
Call Trace:
try_charge+0xcb/0x780
memcg_kmem_charge_memcg+0x28/0x80
memcg_kmem_charge+0x8b/0x1d0
copy_process.part.41+0x1ca/0x2070
_do_fork+0xd7/0x3d0
do_syscall_64+0x5a/0x180
entry_SYSCALL_64_after_hwframe+0x49/0xbe
The problem occurs because get_mem_cgroup_from_current() returns the NULL
pointer if memory controller is disabled. Let's check if this is a case
at the beginning of memcg_kmem_charge() and just return 0 if
mem_cgroup_disabled() returns true. This is how we handle this case in
many other places in the memory controller code.
Link: http://lkml.kernel.org/r/20181029215123.17830-1-guro@fb.com
Fixes: 9b6f7e163cd0 ("mm: rework memcg kernel stack accounting")
Signed-off-by: Roman Gushchin <guro@fb.com>
Reported-by: Mike Galbraith <efault@gmx.de>
Acked-by: Rik van Riel <riel@surriel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Shakeel Butt <shakeelb@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
Remove unused parameter from HASH1 dmas property on stm32mp157c SoC.
Fixes: 1e726a40e067 ("ARM: dts: stm32: Add HASH support on stm32mp157c")
Signed-off-by: Alexandre Torgue <alexandre.torgue@st.com>
[Olof: Bug doesn't cause any harm, so shouldn't need stable backport]
Signed-off-by: Olof Johansson <olof@lixom.net>
|
|
Testing randconfig builds found an instance of a VLA that was
missed when determining that we have removed them all:
arch/arm/plat-orion/mpp.c: In function 'orion_mpp_conf':
arch/arm/plat-orion/mpp.c:31:2: error: ISO C90 forbids variable length array 'mpp_ctrl' [-Werror=vla]
This one is fairly straightforward: we know what all three
callers are, and the maximum length is not very long.
Fixes: 68664695ae57 ("Makefile: Globally enable VLA warning")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Olof Johansson <olof@lixom.net>
|
|
When switching to the new iovec accessors, a negation got subtly
dropped, leading to 9p being remarkably broken (here with kvmtool):
[ 7.430941] VFS: Mounted root (9p filesystem) on device 0:15.
[ 7.432080] devtmpfs: mounted
[ 7.432717] Freeing unused kernel memory: 1344K
[ 7.433658] Run /virt/init as init process
Warning: unable to translate guest address 0x7e00902ff000 to host
Warning: unable to translate guest address 0x7e00902fefc0 to host
Warning: unable to translate guest address 0x7e00902ff000 to host
Warning: unable to translate guest address 0x7e008febef80 to host
Warning: unable to translate guest address 0x7e008febf000 to host
Warning: unable to translate guest address 0x7e008febef00 to host
Warning: unable to translate guest address 0x7e008febf000 to host
[ 7.436376] Kernel panic - not syncing: Requested init /virt/init failed (error -8).
[ 7.437554] CPU: 29 PID: 1 Comm: swapper/0 Not tainted 4.19.0-rc8-02267-g00e23707442a #291
[ 7.439006] Hardware name: linux,dummy-virt (DT)
[ 7.439902] Call trace:
[ 7.440387] dump_backtrace+0x0/0x148
[ 7.441104] show_stack+0x14/0x20
[ 7.441768] dump_stack+0x90/0xb4
[ 7.442425] panic+0x120/0x27c
[ 7.443036] kernel_init+0xa4/0x100
[ 7.443725] ret_from_fork+0x10/0x18
[ 7.444444] SMP: stopping secondary CPUs
[ 7.445391] Kernel Offset: disabled
[ 7.446169] CPU features: 0x0,23000438
[ 7.446974] Memory Limit: none
[ 7.447645] ---[ end Kernel panic - not syncing: Requested init /virt/init failed (error -8). ]---
Restoring the missing "!" brings the guest back to life.
Fixes: 00e23707442a ("iov_iter: Use accessor function")
Reported-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
|
|
The patch "CIFS: Add support for direct I/O read" had
a signed/unsigned mismatch (ssize_t vs. size_t) in the
return from one function. Similar trivial change
in aio_write
Signed-off-by: Long Li <longli@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Reported-by: Julia Lawall <julia.lawall@lip6.fr>
|
|
There is a null check on dst_file->private data which suggests
it can be potentially null. However, before this check, pointer
smb_file_target is derived from dst_file->private and dereferenced
in the call to tlink_tcon, hence there is a potential null pointer
deference.
Fix this by assigning smb_file_target and target_tcon after the
null pointer sanity checks.
Detected by CoverityScan, CID#1475302 ("Dereference before null check")
Fixes: 04b38d601239 ("vfs: pull btrfs clone API to vfs layer")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
With direct read/write functions implemented, add them to file_operations.
Dircet I/O is used under two conditions:
1. When mounting with "cache=none", CIFS uses direct I/O for all user file
data transfer.
2. When opening a file with O_DIRECT, CIFS uses direct I/O for all data
transfer on this file.
Signed-off-by: Long Li <longli@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
|
|
With direct I/O write, user supplied buffers are pinned to the memory and data
are transferred directly from user buffers to the transport layer.
Change in v3: add support for kernel AIO
Change in v4:
Refactor common write code to __cifs_writev for direct and non-direct I/O.
Retry on direct I/O failure.
Signed-off-by: Long Li <longli@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
With direct I/O read, we transfer the data directly from transport layer to
the user data buffer.
Change in v3: add support for kernel AIO
Change in v4:
Refactor common read code to __cifs_readv for direct and non-direct I/O.
Retry on direct I/O failure.
Signed-off-by: Long Li <longli@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
We were missing some structs from MS-FSCC relating to
reparse point handling. Add them to protocol defines
in smb2pdu.h
Signed-off-by: Steve French <stfrench@microsoft.com>
Reviewed-by: Aurelien Aptel <aaptel@suse.com>
|
|
In order to debug complex problems it is often helpful to
have detailed information on the client and server view
of the open file information. Add the ability for root to
view the list of smb3 open files and dump the persistent
handle and other info so that it can be more easily
correlated with server logs.
Sample output from "cat /proc/fs/cifs/open_files"
# Version:1
# Format:
# <tree id> <persistent fid> <flags> <count> <pid> <uid> <filename> <mid>
0x5 0x800000378 0x8000 1 7704 0 some-file 0x14
0xcb903c0c 0x84412e67 0x8000 1 7754 1001 rofile 0x1a6d
0xcb903c0c 0x9526b767 0x8000 1 7720 1000 file 0x1a5b
0xcb903c0c 0x9ce41a21 0x8000 1 7715 0 smallfile 0xd67
Signed-off-by: Steve French <stfrench@microsoft.com>
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
|
|
Some servers (e.g. Azure) do not include a spnego blob in the SMB3
negotiate protocol response, so on kerberos mounts ("sec=krb5")
we can fail, as we expected the server to list its supported
auth types (OIDs in the spnego blob in the negprot response).
Change this so that on krb5 mounts we default to trying krb5 if the
server doesn't list its supported protocol mechanisms.
Signed-off-by: Steve French <stfrench@microsoft.com>
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
CC: Stable <stable@vger.kernel.org>
|
|
In debugging certain scenarios, especially reconnect cases,
it can be helpful to have a dynamic trace point for the
result of tree connect. See sample output below
from a reconnect event. The new event is 'smb3_tcon'
TASK-PID CPU# |||| TIMESTAMP FUNCTION
| | | |||| | |
cifsd-6071 [001] .... 2659.897923: smb3_reconnect: server=localhost current_mid=0xa
kworker/1:1-71 [001] .... 2666.026342: smb3_cmd_done: sid=0x0 tid=0x0 cmd=0 mid=0
kworker/1:1-71 [001] .... 2666.026576: smb3_cmd_err: sid=0xc49e1787 tid=0x0 cmd=1 mid=1 status=0xc0000016 rc=-5
kworker/1:1-71 [001] .... 2666.031677: smb3_cmd_done: sid=0xc49e1787 tid=0x0 cmd=1 mid=2
kworker/1:1-71 [001] .... 2666.031921: smb3_cmd_done: sid=0xc49e1787 tid=0x6e78f05f cmd=3 mid=3
kworker/1:1-71 [001] .... 2666.031923: smb3_tcon: xid=0 sid=0xc49e1787 tid=0x0 unc_name=\\localhost\test rc=0
kworker/1:1-71 [001] .... 2666.032097: smb3_cmd_done: sid=0xc49e1787 tid=0x6e78f05f cmd=11 mid=4
kworker/1:1-71 [001] .... 2666.032265: smb3_cmd_done: sid=0xc49e1787 tid=0x7912332f cmd=3 mid=5
kworker/1:1-71 [001] .... 2666.032266: smb3_tcon: xid=0 sid=0xc49e1787 tid=0x0 unc_name=\\localhost\IPC$ rc=0
kworker/1:1-71 [001] .... 2666.032386: smb3_cmd_done: sid=0xc49e1787 tid=0x7912332f cmd=11 mid=6
Signed-off-by: Steve French <stfrench@microsoft.com>
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
|
|
Trivial fix to a spelling mistake of the error access name EACCESS,
rename to EACCES
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
If the application buffer was too small to fit all the names
we would still count the number of bytes and return this for
listxattr. This would then trigger a BUG in usercopy.c
Fix the computation of the size so that we return -ERANGE
correctly when the buffer is too small.
This fixes the kernel BUG for xfstest generic/377
Signed-off-by: Ronnie Sahlberg <lsahlber@redhat.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Reviewed-by: Aurelien Aptel <aaptel@suse.com>
|
Ulf Hansson
Viresh Kumar
Linus Torvalds
Peter Zijlstra
Michal Hocko
Tetsuo Handa
Michael Schupikov
zhong jiang
Larry Chen
Arnd Bergmann
Changwei Ge
Guozhonghua
Gang He
Andrea Arcangeli
Sam Protsenko
Roman Gushchin
Alexandre Torgue
Marc Zyngier
Steve French
Colin Ian King
Long Li
Ronnie Sahlberg