While I play inhouse patches with much memory pressure on qemu-kvm,
3.14 kernel was randomly crashed. The reason was kernel stack overflow.
When I investigated the problem, the callstack was a little bit deeper
by involve with reclaim functions but not direct reclaim path.
I tried to diet stack size of some functions related with alloc/reclaim
so did a hundred of byte but overflow was't disappeard so that I encounter
overflow by another deeper callstack on reclaim/allocator path.
Of course, we might sweep every sites we have found for reducing
stack usage but I'm not sure how long it saves the world(surely,
lots of developer start to add nice features which will use stack
agains) and if we consider another more complex feature in I/O layer
and/or reclaim path, it might be better to increase stack size(
meanwhile, stack usage on 64bit machine was doubled compared to 32bit
while it have sticked to 8K. Hmm, it's not a fair to me and arm64
already expaned to 16K. )
So, my stupid idea is just let's expand stack size and keep an eye
toward stack consumption on each kernel functions via stacktrace of ftrace.
For example, we can have a bar like that each funcion shouldn't exceed 200K
and emit the warning when some function consumes more in runtime.
Of course, it could make false positive but at least, it could make a
chance to think over it.
I guess this topic was discussed several time so there might be
strong reason not to increase kernel stack size on x86_64, for me not
knowing so Ccing x86_64 maintainers, other MM guys and virtio
maintainers.
Here's an example call trace using up the kernel stack:
Depth Size Location (51 entries)
----- ---- --------
0) 7696 16 lookup_address
1) 7680 16 _lookup_address_cpa.isra.3
2) 7664 24 __change_page_attr_set_clr
3) 7640 392 kernel_map_pages
4) 7248 256 get_page_from_freelist
5) 6992 352 __alloc_pages_nodemask
6) 6640 8 alloc_pages_current
7) 6632 168 new_slab
8) 6464 8 __slab_alloc
9) 6456 80 __kmalloc
10) 6376 376 vring_add_indirect
11) 6000 144 virtqueue_add_sgs
12) 5856 288 __virtblk_add_req
13) 5568 96 virtio_queue_rq
14) 5472 128 __blk_mq_run_hw_queue
15) 5344 16 blk_mq_run_hw_queue
16) 5328 96 blk_mq_insert_requests
17) 5232 112 blk_mq_flush_plug_list
18) 5120 112 blk_flush_plug_list
19) 5008 64 io_schedule_timeout
20) 4944 128 mempool_alloc
21) 4816 96 bio_alloc_bioset
22) 4720 48 get_swap_bio
23) 4672 160 __swap_writepage
24) 4512 32 swap_writepage
25) 4480 320 shrink_page_list
26) 4160 208 shrink_inactive_list
27) 3952 304 shrink_lruvec
28) 3648 80 shrink_zone
29) 3568 128 do_try_to_free_pages
30) 3440 208 try_to_free_pages
31) 3232 352 __alloc_pages_nodemask
32) 2880 8 alloc_pages_current
33) 2872 200 __page_cache_alloc
34) 2672 80 find_or_create_page
35) 2592 80 ext4_mb_load_buddy
36) 2512 176 ext4_mb_regular_allocator
37) 2336 128 ext4_mb_new_blocks
38) 2208 256 ext4_ext_map_blocks
39) 1952 160 ext4_map_blocks
40) 1792 384 ext4_writepages
41) 1408 16 do_writepages
42) 1392 96 __writeback_single_inode
43) 1296 176 writeback_sb_inodes
44) 1120 80 __writeback_inodes_wb
45) 1040 160 wb_writeback
46) 880 208 bdi_writeback_workfn
47) 672 144 process_one_work
48) 528 112 worker_thread
49) 416 240 kthread
50) 176 176 ret_from_fork
[ Note: the problem is exacerbated by certain gcc versions that seem to
generate much bigger stack frames due to apparently bad coalescing of
temporaries and generating too many spills. Rusty saw gcc-4.6.4 using
35% more stack on the virtio path than 4.8.2 does, for example.
Minchan not only uses such a bad gcc version (4.6.3 in his case), but
some of the stack use is due to debugging (CONFIG_DEBUG_PAGEALLOC is
what causes that kernel_map_pages() frame, for example). But we're
clearly getting too close.
The VM code also seems to have excessive stack frames partly for the
same compiler reason, triggered by excessive inlining and lots of
function arguments.
We need to improve on our stack use, but in the meantime let's do this
simple stack increase too. Unlike most earlier reports, there is
nothing simple that stands out as being really horribly wrong here,
apart from the fact that the stack frames are just bigger than they
should need to be. - Linus ]
Signed-off-by: Minchan Kim <minchan@kernel.org>
Cc: Peter Anvin <hpa@zytor.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Dave Jones <davej@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rik van Riel <riel@redhat.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Michael S Tsirkin <mst@redhat.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: PJ Waskiewicz <pjwaskiewicz@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Since commit 166aaf39 ("ARM: 8029/1: mcpm: Rename the power_down_finish()
functions to be less confusing") changed the name of power_down_finish to
wait_for_cpu_powerdown, so use new member name wait_for_cpu_powerdown.
Reported-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
The exynos5800 is very similar to exynos5420. We can re-use
the existing MCPM support for exynos5800 for secondary boot
-up and switching.
Signed-off-by: Abhilash Kesavan <a.kesavan@samsung.com>
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
This makes it possible to enable the Exynos platform as part of a
multiplatform kernel.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
Instead of repeating the Kconfig entries for every SoC,
move them under ARCH_EXYNOS3, 4 and 5 and move the entries
common to 3, 4 and 5 under ARCH_EXYNOS.
Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
This patch fix the offset of CPU boot address and don't
need to send smc call of SMC_CMD_CPU1BOOT command for
secondary CPU boot because Exynos3250 removes WFE in
secure mode.
Signed-off-by: Chanwoo Choi <cw00.choi@samsung.com>
Acked-by: Kyungmin Park <kyungmin.park@samsung.com>
Reviewed-by: Tomasz Figa <t.figa@samsung.com>
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
This patch add Exynos3250's SoC ID. Exynos 3250 is SoC that
is based on the 32-bit RISC processor for Smartphone.
Exynos3250 uses Cortex-A7 dual cores and has a target speed
of 1.0GHz.
Signed-off-by: Chanwoo Choi <cw00.choi@samsung.com>
Acked-by: Kyungmin Park <kyungmin.park@samsung.com>
Reviewed-by: Tomasz Figa <t.figa@samsung.com>
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
Exynos5800 is an octa core SoC which is based on the 5420
platform. This patch adds the basic support for it in the
mach-exynos.
Signed-off-by: Arun Kumar K <arun.kk@samsung.com>
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
This patch add basic arch side support for exynos5260 SoC.
Note that this is required to enable build for clock driver.
Signed-off-by: Pankaj Dubey <pankaj.dubey@samsung.com>
Signed-off-by: Rahul Sharma <rahul.sharma@samsung.com>
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
The vbus-supply property is wrongly updated in the
usbdrd node instead of the usbdrd_phy node. This patch
fixes the same.
Signed-off-by: Arun Kumar K <arun.kk@samsung.com>
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
The vbus-supply property is wrongly updated in the
usbdrd node instead of the usbdrd_phy node. This patch
fixes the same.
Signed-off-by: Arun Kumar K <arun.kk@samsung.com>
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
Most of the nodes of exynos5420 remains same for exynos5800.
So the exynos5420.dtsi is included in exynos5800 and the changed
node properties will be overriden.
Signed-off-by: Arun Kumar K <arun.kk@samsung.com>
Reviewed-by: Tomasz Figa <t.figa@samsung.com>
Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
The patch adds the dts file for xyref5260 board which
is based on exynos5260 SoC.
Signed-off-by: Rahul Sharma <rahul.sharma@samsung.com>
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
Key code macros improve readability on exnos4210-origen,
exynos4412-origen and exynos5250-arndale boards.
Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
[kgene.kim@samsung.com: squashed similar two patches]
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
Enabled RTC and WDT nodes on exynos4210-origen and
exynos4412-origen boards.
Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
[kgene.kim@samsung.com: squashed similar two patches]
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
Call pcie_bus_configure_settings() on ARM, like for other platforms.
pcie_bus_configure_settings() makes sure the MPS across the bus is uniform
and provides the ability to tune the MRSS and MPS to higher performance
values. This is particularly important for embedded where there is no
firmware to program these PCIe settings for the OS.
Signed-off-by: Murali Karicheri <m-karicheri2@ti.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
CC: Russell King <linux@arm.linux.org.uk>
CC: Arnd Bergmann <arnd@arndb.de>
CC: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
CC: Santosh Shilimkar <santosh.shilimkar@ti.com>
On platforms implementing CPU power management, the CPUidle subsystem
can allow CPUs to enter idle states where local timers logic is lost on power
down. To keep the software timers functional the kernel relies on an
always-on broadcast timer to be present in the platform to relay the
interrupt signalling the timer expiries.
For platforms implementing CPU core gating that do not implement an always-on
HW timer or implement it in a broken way, this patch adds code to initialize
the kernel hrtimer based clock event device upon boot (which can be chosen as
tick broadcast device by the kernel).
It relies on a dynamically chosen CPU to be always powered-up. This CPU then
relays the timer interrupt to CPUs in deep-idle states through its HW local
timer device.
Having a CPU always-on has implications on power management platform
capabilities and makes CPUidle suboptimal, since at least a CPU is kept
always in a shallow idle state by the kernel to relay timer interrupts,
but at least leaves the kernel with a functional system with some working
power management capabilities.
The hrtimer based clock event device is unconditionally registered, but
has the lowest possible rating such that any broadcast-capable HW clock
event device present will be chosen in preference as the tick broadcast
device.
Reviewed-by: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Acked-by: Will Deacon <will.deacon@arm.com>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Merge "Renesas ARM Based SoC Fixes for v3.16" from Simon Horman:
This corrects a bug that will be introduced in v3.15.
The bug causes audio playback to fail on the Armadillo800 EVA board.
* tag 'renesas-fixes-for-v3.16' of git://git.kernel.org/pub/scm/linux/kernel/git/horms/renesas:
ARM: shmobile: armadillo800eva: fixup HDMI sound flags setting
Signed-off-by: Olof Johansson <olof@lixom.net>
Update to commit 9c9b415c50 [MIPS:
Reimplement get_cycles().]
On systems were for whatever reasons we can't use the cycle counter, fall
back to the c0_random register as an entropy source. It has however a
very small range that makes it suitable for random_get_entropy only and
not get_cycles.
This optimised version compiles to 8 instructions in the fast path even in
the worst case of all the conditions to check being variable (including a
MFC0 move delay slot that is only required for very old processors):
828: 8cf90000 lw t9,0(a3)
828: R_MIPS_LO16 jiffies
82c: 40057800 mfc0 a1,c0_prid
830: 3c0200ff lui v0,0xff
834: 00a21024 and v0,a1,v0
838: 1040007d beqz v0,a30 <add_interrupt_randomness+0x22c>
83c: 3c030000 lui v1,0x0
83c: R_MIPS_HI16 cpu_data
840: 40024800 mfc0 v0,c0_count
844: 00000000 nop
848: 00409021 move s2,v0
84c: 8ce20000 lw v0,0(a3)
84c: R_MIPS_LO16 jiffies
On most targets the sequence will be shorter and on some it will reduce to
a single `MFC0 <reg>,c0_count', as all MIPS architecture (i.e. non-legacy
MIPS) processors require the CP0 Count register to be present.
The only known exception that reports MIPS architecture compliance, but
contrary to that lacks CP0 Count is the Ingenic JZ4740 thingy. For broken
platforms like that this code requires cpu_has_counter to be hardcoded to
0 (i.e. no variable setting is permitted) so as not to penalise all the
other good platforms out there.
The asm barrier is required so that the compiler does not pull any
potentially costly (cold cache!) `cpu_data' variable access into the fast
path.
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: John Crispin <blogic@openwrt.org>
Cc: Andrew McGregor <andrewmcgr@gmail.com>
Cc: Dave Taht <dave.taht@bufferbloat.net>
Cc: Felix Fietkau <nbd@nbd.name>
Cc: Simon Kelley <simon@thekelleys.org.uk>
Cc: Jim Gettys <jg@freedesktop.org>
Cc: David Daney <ddaney@caviumnetworks.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6702/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
