This is an improved straw bucket that correctly avoids any data movement
between items A and B when neither A nor B's weights are changed. Said
differently, if we adjust the weight of item C (including adding it anew
or removing it completely), we will only see inputs move to or from C,
never between other items in the bucket.
Notably, there is not intermediate scaling factor that needs to be
calculated. The mapping function is a simple function of the item weights.
The below commits were squashed together into this one (mostly to avoid
adding and then yanking a ~6000 lines worth of crush_ln_table):
- crush: add a straw2 bucket type
- crush: add crush_ln to calculate nature log efficently
- crush: improve straw2 adjustment slightly
- crush: change crush_ln to provide 32 more digits
- crush: fix crush_get_bucket_item_weight and bucket destroy for straw2
- crush/mapper: fix divide-by-0 in straw2
(with div64_s64() for draw = ln / w and INT64_MIN -> S64_MIN - need
to create a proper compat.h in ceph.git)
Reflects ceph.git commits 242293c908e923d474910f2b8203fa3b41eb5a53,
32a1ead92efcd351822d22a5fc37d159c65c1338,
6289912418c4a3597a11778bcf29ed5415117ad9,
35fcb04e2945717cf5cfe150b9fa89cb3d2303a1,
6445d9ee7290938de1e4ee9563912a6ab6d8ee5f,
b5921d55d16796e12d66ad2c4add7305f9ce2353.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Crush temporary buffers are allocated as per replica size configured
by the user. When there are more final osds (to be selected as per
rule) than the replicas, buffer overlaps and it causes crash. Now, it
ensures that at most num-rep osds are selected even if more number of
osds are allowed by the rule.
Reflects ceph.git commits 6b4d1aa99718e3b367496326c1e64551330fabc0,
234b066ba04976783d15ff2abc3e81b6cc06fb10.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Rather small this time:
- a fix for a nasty bug with virtual IRQ injection
- a fix for irqfd
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1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=Cfx3
-----END PGP SIGNATURE-----
Merge tag 'kvm-arm-for-4.1-take2' of git://git.kernel.org/pub/scm/linux/kernel/git/kvmarm/kvmarm into kvm-master
KVM/ARM changes for v4.1, take #2:
Rather small this time:
- a fix for a nasty bug with virtual IRQ injection
- a fix for irqfd
When userland injects a SPI via the KVM_IRQ_LINE ioctl we currently
only check it against a fixed limit, which historically is set
to 127. With the new dynamic IRQ allocation the effective limit may
actually be smaller (64).
So when now a malicious or buggy userland injects a SPI in that
range, we spill over on our VGIC bitmaps and bytemaps memory.
I could trigger a host kernel NULL pointer dereference with current
mainline by injecting some bogus IRQ number from a hacked kvmtool:
-----------------
....
DEBUG: kvm_vgic_inject_irq(kvm, cpu=0, irq=114, level=1)
DEBUG: vgic_update_irq_pending(kvm, cpu=0, irq=114, level=1)
DEBUG: IRQ #114 still in the game, writing to bytemap now...
Unable to handle kernel NULL pointer dereference at virtual address 00000000
pgd = ffffffc07652e000
[00000000] *pgd=00000000f658b003, *pud=00000000f658b003, *pmd=0000000000000000
Internal error: Oops: 96000006 [#1] PREEMPT SMP
Modules linked in:
CPU: 1 PID: 1053 Comm: lkvm-msi-irqinj Not tainted 4.0.0-rc7+ #3027
Hardware name: FVP Base (DT)
task: ffffffc0774e9680 ti: ffffffc0765a8000 task.ti: ffffffc0765a8000
PC is at kvm_vgic_inject_irq+0x234/0x310
LR is at kvm_vgic_inject_irq+0x30c/0x310
pc : [<ffffffc0000ae0a8>] lr : [<ffffffc0000ae180>] pstate: 80000145
.....
So this patch fixes this by checking the SPI number against the
actual limit. Also we remove the former legacy hard limit of
127 in the ioctl code.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Christoffer Dall <christoffer.dall@linaro.org>
CC: <stable@vger.kernel.org> # 4.0, 3.19, 3.18
[maz: wrap KVM_ARM_IRQ_GIC_MAX with #ifndef __KERNEL__,
as suggested by Christopher Covington]
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
irqfd/arm curently does not support routing. kvm_irq_map_gsi is
supposed to return all the routing entries associated with the
provided gsi and return the number of those entries. We should
return 0 at this point.
Signed-off-by: Eric Auger <eric.auger@linaro.org>
Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
My Pengutronix address is not valid anymore, redirect people to the Pengutronix
kernel team.
Reported-by: Harald Geyer <harald@ccbib.org>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Acked-by: Robert Schwebel <r.schwebel@pengutronix.de>
Signed-off-by: Wim Van Sebroeck <wim@iguana.be>
My Pengutronix address is not valid anymore, redirect people to the Pengutronix
kernel team.
Reported-by: Harald Geyer <harald@ccbib.org>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Acked-by: Robert Schwebel <r.schwebel@pengutronix.de>
Signed-off-by: Wim Van Sebroeck <wim@iguana.be>
Use fixed length string for register names. This saves 416 bytes
in text size.
Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Reviewed-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Wim Van Sebroeck <wim@iguana.be>
Fix some trivial coding style issues to reduce noise from static analyzers.
Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Reviewed-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Wim Van Sebroeck <wim@iguana.be>
Convert OCTEON watchdog to WATCHDOG_CORE API. This enables support
for multiple watchdogs on OCTEON boards.
Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Reviewed-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Wim Van Sebroeck <wim@iguana.be>
Remove Kconfig dependency and enable driver for
all ARCHs.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Wim Van Sebroeck <wim@iguana.be>
The watchdog has been reworked to use the same DT node as the timer.
This change is updating the device tree doc accordingly.
Signed-off-by: Mathieu Olivari <mathieu@codeaurora.org>
Acked-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Wim Van Sebroeck <wim@iguana.be>
Add the watchdog related entries to the Krait Processor Sub-system
(KPSS) timer IPQ8064 devicetree section. Also, add a fixed-clock
description of SLEEP_CLK, which will do for now.
Signed-off-by: Josh Cartwright <joshc@codeaurora.org>
Signed-off-by: Mathieu Olivari <mathieu@codeaurora.org>
Reviewed-by: Stephen Boyd <sboyd@codeaurora.org>
Acked-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Wim Van Sebroeck <wim@iguana.be>
MSM watchdog configuration happens in the same register block as the
timer, so we'll use the same binding as the existing timer.
The qcom-wdt will now be probed when devicetree has an entry compatible
with "qcom,kpss-timer" or "qcom-scss-timer".
Signed-off-by: Mathieu Olivari <mathieu@codeaurora.org>
Reviewed-by: Stephen Boyd <sboyd@codeaurora.org>
Acked-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Wim Van Sebroeck <wim@iguana.be>
The seq_printf return value, because it's frequently misused,
will eventually be converted to void.
See: commit 1f33c41c03 ("seq_file: Rename seq_overflow() to
seq_has_overflowed() and make public")
Signed-off-by: Joe Perches <joe@perches.com>
Acked-by: Guenter Roeck <linux~roeck-us.net>
Signed-off-by: Wim Van Sebroeck <wim@iguana.be>
Currently an allyesconfig build [gcc-4.9.1] can generate the following:
WARNING: vmlinux.o(.text.unlikely+0x3864): Section mismatch in
reference from the function cpumask_empty.constprop.3() to the
variable .init.data:nmi_ipi_mask
which comes from the cpumask_empty usage in arch/x86/kernel/nmi_selftest.c.
Normally we would not see a symbol entry for cpumask_empty since it is:
static inline bool cpumask_empty(const struct cpumask *srcp)
however in this case, the variant of the symbol gets emitted when GCC does
constant propagation optimization.
Fix things up so that any locally optimized constprop variants don't warn
when accessing variables that live in the __init sections.
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Currently the match() function supports a leading * to match any
prefix and a trailing * to match any suffix. However there currently
is not a combination of both that can be used to target matches of
whole families of functions that share a common substring.
Here we expand the *foo and foo* match to also support *foo* with
the goal of targeting compiler generated symbol names that contain
strings like ".constprop." and ".isra."
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Trying to match the SHT_NUL section isn't useful and causes build failures
on parisc and mn10300 since the addition of section strict white-listing
and __ex_table sanitizing.
Signed-off-by: Quentin Casasnovas <quentin.casasnovas@oracle.com>
Reported-by: Guenter Roeck <linux@roeck-us.net>
Fixes: 050e57fd59 ("modpost: add strict white-listing when referencing....")
Fixes: 52dc0595d5 ("modpost: handle relocations mismatch in __ex_table.")
Tested-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
As Guenter pointed out, we were never really calculating the extable entry
size because the pointer arithmetic was simply wrong. We want to check
we're handling the second relocation in __ex_table to infer an entry size,
but we were using (void*) pointers instead of Elf_Rel[a]* ones.
This fixes the problem by moving that check in the caller (since we can
deal with different types of relocations) and add is_second_extable_reloc()
to make the whole thing more readable.
Signed-off-by: Quentin Casasnovas <quentin.casasnovas@oracle.com>
Reported-by: Guenter Roeck <linux@roeck-us.net>
CC: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
As Guenter pointed out, we want to assert that extable_entry_size has been
discovered and not the other way around. Moreover, this sanity check is
only valid when we're not dealing with the first relocation in __ex_table,
since we have not discovered the extable entry size at that point.
This was leading to a divide-by-zero on some architectures and make the
build fail.
Signed-off-by: Quentin Casasnovas <quentin.casasnovas@oracle.com>
Reported-by: Guenter Roeck <linux@roeck-us.net>
CC: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
52dc0595d5 introduced OTHER_TEXT_SECTIONS for identifying what
sections could validly have __ex_table entries. Unfortunately, it
wasn't tested with -ffunction-sections, which some architectures
use.
Reported-by: kbuild test robot <fengguang.wu@intel.com>
Cc: Quentin Casasnovas <quentin.casasnovas@oracle.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
32-bit and 64-bit ARM use these sections to store executable code, so
they must be whitelisted in modpost's table of valid text sections.
Signed-off-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
DW_DMAC_CORE is slected by PCI or Platform driver, so this symbol shouldn't
be user selectable, so remove the prompt
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Pull sparc fixes from David Miller:
1) ldc_alloc_exp_dring() can be called from softints, so use
GFP_ATOMIC. From Sowmini Varadhan.
2) Some minor warning/build fixups for the new iommu-common code on
certain archs and with certain debug options enabled. Also from
Sowmini Varadhan.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc:
sparc: Use GFP_ATOMIC in ldc_alloc_exp_dring() as it can be called in softirq context
sparc64: Use M7 PMC write on all chips T4 and onward.
iommu-common: rename iommu_pool_hash to iommu_hash_common
iommu-common: fix x86_64 compiler warnings
Pull networking fixes from David Miller:
"Just a few fixes trickling in at this point.
1) If we see an attached socket on an skb in the ipv4 forwarding path,
bail. This can happen due to races with FIB rule addition, and
deletion, and we should just drop such frames. From Sebastian
Pöhn.
2) pppoe receive should only accept packets destined for this hosts's
MAC address. From Joakim Tjernlund.
3) Handle checksum unwrapping properly in ppp receive properly when
it's encapsulated in UDP in some way, fix from Tom Herbert.
4) Fix some bugs in mv88e6xxx DSA driver resulting from the conversion
from register offset constants to mnenomic macros. From Vivien
Didelot.
5) Fix handling of HCA max message size in mlx4 adapters, from Eran
Ben ELisha"
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net:
net/mlx4_core: Fix reading HCA max message size in mlx4_QUERY_DEV_CAP
tcp: add memory barriers to write space paths
altera tse: Error-Bit on tx-avalon-stream always set.
net: dsa: mv88e6xxx: use PORT_DEFAULT_VLAN
net: dsa: mv88e6xxx: fix setup of port control 1
ppp: call skb_checksum_complete_unset in ppp_receive_frame
net: add skb_checksum_complete_unset
pppoe: Lacks DST MAC address check
ip_forward: Drop frames with attached skb->sk
.config files that do not have MFD_SYSCON set.
Omaps now have a dependency to MFD_SYSCON for system control
module generic register area and some clocks with the changes
done in omap-for-v4.1/prcm-dts branch.
This can be pulled on top of omap-for-v4.1/prcm-dts, or into
fixes for v4.1.
We already do have a slight MFD_SYSCON dependency for
REGULATOR_PBIAS for dual voltage MMC cards on the first MMC
bus for many devices, so from that point of view this can
also be merged separately from omap-for-v4.1/prcm-dts.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1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=4WLp
-----END PGP SIGNATURE-----
Merge tag 'omap-for-v4.1/prcm-dts-mfd-syscon-fix' of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap into next/late
Merge "urgent omap boot fix for v4.1 if MFD_SYSCON is not set" from Tony
Lindgren:
Urgent pull request for v4.1 to booting for custom kernel
.config files that do not have MFD_SYSCON set.
Omaps now have a dependency to MFD_SYSCON for system control
module generic register area and some clocks with the changes
done in omap-for-v4.1/prcm-dts branch.
This can be pulled on top of omap-for-v4.1/prcm-dts, or into
fixes for v4.1.
We already do have a slight MFD_SYSCON dependency for
REGULATOR_PBIAS for dual voltage MMC cards on the first MMC
bus for many devices, so from that point of view this can
also be merged separately from omap-for-v4.1/prcm-dts.
* tag 'omap-for-v4.1/prcm-dts-mfd-syscon-fix' of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap:
ARM: OMAP2+: Fix booting with configs that don't have MFD_SYSCON
Signed-off-by: Olof Johansson <olof@lixom.net>
Use list_for_each_entry_safe for iterating because handler may be freed
in the loop.
BUG: unable to handle kernel NULL pointer dereference at 000000000000002c
IP: [<ffffffff814d69c8>] acpi_ec_put_query_handler+0x7/0x1a
Call Trace:
acpi_ec_remove_query_handler+0x87/0x97
acpi_smbus_hc_remove+0x2a/0x44 [sbshc]
acpi_device_remove+0x7b/0x9a
__device_release_driver+0x7e/0x110
driver_detach+0xb0/0xc0
bus_remove_driver+0x54/0xe0
driver_unregister+0x2b/0x60
acpi_bus_unregister_driver+0x10/0x12
acpi_smb_hc_driver_exit+0x10/0x12 [sbshc]
SyS_delete_module+0x1b8/0x210
system_call_fastpath+0x12/0x6a
Signed-off-by: Chris Bainbridge <chris.bainbridge@gmail.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Pull parisc fixes from Helge Deller:
"The patch by Guenter Roeck fixes the build on parisc which got broken
because of commit f24ffde432 ("parisc: expose number of page table
levels on Kconfig level") and the patch from Matthew Wilcox converts
our code to use the generic scatterlist.h header file"
* 'parisc-4.1-1' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux:
parisc: Replace PT_NLEVELS with CONFIG_PGTABLE_LEVELS
parisc: Eliminate sg_virt_addr() and private scatterlist.h
When array is degraded, read data landed on failed drives will result in
reading rest of data in a stripe. So a single sequential read would
result in same data being read twice.
This patch is to avoid chunk aligned read for degraded array. The
downside is to involve stripe cache which means associated CPU overhead
and extra memory copy.
Test Results:
Following test are done on a enterprise storage node with Seagate 6T SAS
drives and Xeon E5-2648L CPU (10 cores, 1.9Ghz), 10 disks MD RAID6 8+2,
chunk size 128 KiB.
I use FIO, using direct-io with various bs size, enough queue depth,
tested sequential and 100% random read against 3 array config:
1) optimal, as baseline;
2) degraded;
3) degraded with this patch.
Kernel version is 4.0-rc3.
Each individual test I only did once so there might be some variations,
but we just focus on big trend.
Sequential Read:
bs=(KiB) optimal(MiB/s) degraded(MiB/s) degraded-with-patch (MiB/s)
1024 1608 656 995
512 1624 710 956
256 1635 728 980
128 1636 771 983
64 1612 1119 1000
32 1580 1420 1004
16 1368 688 986
8 768 647 953
4 411 413 850
Random Read:
bs=(KiB) optimal(IOPS) degraded(IOPS) degraded-with-patch (IOPS)
1024 163 160 156
512 274 273 272
256 426 428 424
128 576 592 591
64 726 724 726
32 849 848 837
16 900 970 971
8 927 940 929
4 948 940 955
Some notes:
* In sequential + optimal, as bs size getting smaller, the FIO thread
become CPU bound.
* In sequential + degraded, there's big increase when bs is 64K and
32K, I don't have explanation.
* In sequential + degraded-with-patch, the MD thread mostly become CPU
bound.
If you want to we can discuss specific data point in those data. But in
general it seems with this patch, we have more predictable and in most
cases significant better sequential read performance when array is
degraded, and almost no noticeable impact on random read.
Performance is a complicated thing, the patch works well for this
particular configuration, but may not be universal. For example I
imagine testing on all SSD array may have very different result. But I
personally think in most cases IO bandwidth is more scarce resource than
CPU.
Signed-off-by: Eric Mei <eric.mei@seagate.com>
Signed-off-by: NeilBrown <neilb@suse.de>
The default setting of 256 stripe_heads is probably
much too small for many configurations. So it is best to make it
auto-configure.
Shrinking the cache under memory pressure is easy. The only
interesting part here is that we put a fairly high cost
('seeks') on shrinking the cache as the cost is greater than
just having to read more data, it reduces parallelism.
Growing the cache on demand needs to be done carefully. If we allow
fast growth, that can upset memory balance as lots of dirty memory can
quickly turn into lots of memory queued in the stripe_cache.
It is important for the raid5 block device to appear congested to
allow write-throttling to work.
So we only add stripes slowly. We set a flag when an allocation
fails because all stripes are in use, allocate at a convenient
time when that flag is set, and don't allow it to be set again
until at least one stripe_head has been released for re-use.
This means that a spurt of requests will only cause one stripe_head
to be allocated, but a steady stream of requests will slowly
increase the cache size - until memory pressure puts it back again.
It could take hours to reach a steady state.
The value written to, and displayed in, stripe_cache_size is
used as a minimum. The cache can grow above this and shrink back
down to it. The actual size is not directly visible, though it can
be deduced to some extent by watching stripe_cache_active.
Signed-off-by: NeilBrown <neilb@suse.de>
Rather than adjusting max_nr_stripes whenever {grow,drop}_one_stripe()
succeeds, do it inside the functions.
Also choose the correct hash to handle next inside the functions.
This removes duplication and will help with future new uses of
{grow,drop}_one_stripe.
This also fixes a minor bug where the "md/raid:%md: allocate XXkB"
message always said "0kB".
Signed-off-by: NeilBrown <neilb@suse.de>
Depending on the available coding we allow optimized rmw logic for write
operations. To support easier testing this patch allows manual control
of the rmw/rcw descision through the interface /sys/block/mdX/md/rmw_level.
The configuration can handle three levels of control.
rmw_level=0: Disable rmw for all RAID types. Hardware assisted P/Q
calculation has no implementation path yet to factor in/out chunks of
a syndrome. Enforcing this level can be benefical for slow CPUs with
hardware syndrome support and fast SSDs.
rmw_level=1: Estimate rmw IOs and rcw IOs. Execute rmw only if we will
save IOs. This equals the "old" unpatched behaviour and will be the
default.
rmw_level=2: Execute rmw even if calculated IOs for rmw and rcw are
equal. We might have higher CPU consumption because of calculating the
parity twice but it can be benefical otherwise. E.g. RAID4 with fast
dedicated parity disk/SSD. The option is implemented just to be
forward-looking and will ONLY work with this patch!
Signed-off-by: Markus Stockhausen <stockhausen@collogia.de>
Signed-off-by: NeilBrown <neilb@suse.de>
Glue it altogehter. The raid6 rmw path should work the same as the
already existing raid5 logic. So emulate the prexor handling/flags
and split functions as needed.
1) Enable xor_syndrome() in the async layer.
2) Split ops_run_prexor() into RAID4/5 and RAID6 logic. Xor the syndrome
at the start of a rmw run as we did it before for the single parity.
3) Take care of rmw run in ops_run_reconstruct6(). Again process only
the changed pages to get syndrome back into sync.
4) Enhance set_syndrome_sources() to fill NULL pages if we are in a rmw
run. The lower layers will calculate start & end pages from that and
call the xor_syndrome() correspondingly.
5) Adapt the several places where we ignored Q handling up to now.
Performance numbers for a single E5630 system with a mix of 10 7200k
desktop/server disks. 300 seconds random write with 8 threads onto a
3,2TB (10*400GB) RAID6 64K chunk without spare (group_thread_cnt=4)
bsize rmw_level=1 rmw_level=0 rmw_level=1 rmw_level=0
skip_copy=1 skip_copy=1 skip_copy=0 skip_copy=0
4K 115 KB/s 141 KB/s 165 KB/s 140 KB/s
8K 225 KB/s 275 KB/s 324 KB/s 274 KB/s
16K 434 KB/s 536 KB/s 640 KB/s 534 KB/s
32K 751 KB/s 1,051 KB/s 1,234 KB/s 1,045 KB/s
64K 1,339 KB/s 1,958 KB/s 2,282 KB/s 1,962 KB/s
128K 2,673 KB/s 3,862 KB/s 4,113 KB/s 3,898 KB/s
256K 7,685 KB/s 7,539 KB/s 7,557 KB/s 7,638 KB/s
512K 19,556 KB/s 19,558 KB/s 19,652 KB/s 19,688 Kb/s
Signed-off-by: Markus Stockhausen <stockhausen@collogia.de>
Signed-off-by: NeilBrown <neilb@suse.de>
The second and (last) optimized XOR syndrome calculation. This version
supports right and left side optimization. All CPUs with architecture
older than Haswell will benefit from it.
It should be noted that SSE2 movntdq kills performance for memory areas
that are read and written simultaneously in chunks smaller than cache
line size. So use movdqa instead for P/Q writes in sse21 and sse22 XOR
functions.
Signed-off-by: Markus Stockhausen <stockhausen@collogia.de>
Signed-off-by: NeilBrown <neilb@suse.de>
Start the algorithms with the very basic one. It is left and right
optimized. That means we can avoid all calculations for unneeded pages
above the right stop offset. For pages below the left start offset we
still need the syndrome multiplication but without reading data pages.
Signed-off-by: Markus Stockhausen <stockhausen@collogia.de>
Signed-off-by: NeilBrown <neilb@suse.de>
It is always helpful to have a test tool in place if we implement
new data critical algorithms. So add some test routines to the raid6
checker that can prove if the new xor_syndrome() works as expected.
Run through all permutations of start/stop pages per algorithm and
simulate a xor_syndrome() assisted rmw run. After each rmw check if
the recovery algorithm still confirms that the stripe is fine.
Signed-off-by: Markus Stockhausen <stockhausen@collogia.de>
Signed-off-by: NeilBrown <neilb@suse.de>
v3: s-o-b comment, explanation of performance and descision for
the start/stop implementation
Implementing rmw functionality for RAID6 requires optimized syndrome
calculation. Up to now we can only generate a complete syndrome. The
target P/Q pages are always overwritten. With this patch we provide
a framework for inplace P/Q modification. In the first place simply
fill those functions with NULL values.
xor_syndrome() has two additional parameters: start & stop. These
will indicate the first and last page that are changing during a
rmw run. That makes it possible to avoid several unneccessary loops
and speed up calculation. The caller needs to implement the following
logic to make the functions work.
1) xor_syndrome(disks, start, stop, ...): "Remove" all data of source
blocks inside P/Q between (and including) start and end.
2) modify any block with start <= block <= stop
3) xor_syndrome(disks, start, stop, ...): "Reinsert" all data of
source blocks into P/Q between (and including) start and end.
Pages between start and stop that won't be changed should be filled
with a pointer to the kernel zero page. The reasons for not taking NULL
pages are:
1) Algorithms cross the whole source data line by line. Thus avoid
additional branches.
2) Having a NULL page avoids calculating the XOR P parity but still
need calulation steps for the Q parity. Depending on the algorithm
unrolling that might be only a difference of 2 instructions per loop.
The benchmark numbers of the gen_syndrome() functions are displayed in
the kernel log. Do the same for the xor_syndrome() functions. This
will help to analyze performance problems and give an rough estimate
how well the algorithm works. The choice of the fastest algorithm will
still depend on the gen_syndrome() performance.
With the start/stop page implementation the speed can vary a lot in real
life. E.g. a change of page 0 & page 15 on a stripe will be harder to
compute than the case where page 0 & page 1 are XOR candidates. To be not
to enthusiatic about the expected speeds we will run a worse case test
that simulates a change on the upper half of the stripe. So we do:
1) calculation of P/Q for the upper pages
2) continuation of Q for the lower (empty) pages
Signed-off-by: Markus Stockhausen <stockhausen@collogia.de>
Signed-off-by: NeilBrown <neilb@suse.de>
expansion/resync can grab a stripe when the stripe is in batch list. Since all
stripes in batch list must be in the same state, we can't allow some stripes
run into expansion/resync. So we delay expansion/resync for stripe in batch
list.
Signed-off-by: Shaohua Li <shli@fusionio.com>
Signed-off-by: NeilBrown <neilb@suse.de>
If io error happens in any stripe of a batch list, the batch list will be
split, then normal process will run for the stripes in the list.
Signed-off-by: Shaohua Li <shli@fusionio.com>
Signed-off-by: NeilBrown <neilb@suse.de>
stripe cache is 4k size. Even adjacent full stripe writes are handled in 4k
unit. Idealy we should use big size for adjacent full stripe writes. Bigger
stripe cache size means less stripes runing in the state machine so can reduce
cpu overhead. And also bigger size can cause bigger IO size dispatched to under
layer disks.
With below patch, we will automatically batch adjacent full stripe write
together. Such stripes will be added to the batch list. Only the first stripe
of the list will be put to handle_list and so run handle_stripe(). Some steps
of handle_stripe() are extended to cover all stripes of the list, including
ops_run_io, ops_run_biodrain and so on. With this patch, we have less stripes
running in handle_stripe() and we send IO of whole stripe list together to
increase IO size.
Stripes added to a batch list have some limitations. A batch list can only
include full stripe write and can't cross chunk boundary to make sure stripes
have the same parity disks. Stripes in a batch list must be in the same state
(no written, toread and so on). If a stripe is in a batch list, all new
read/write to add_stripe_bio will be blocked to overlap conflict till the batch
list is handled. The limitations will make sure stripes in a batch list be in
exactly the same state in the life circly.
I did test running 160k randwrite in a RAID5 array with 32k chunk size and 6
PCIe SSD. This patch improves around 30% performance and IO size to under layer
disk is exactly 32k. I also run a 4k randwrite test in the same array to make
sure the performance isn't changed with the patch.
Signed-off-by: Shaohua Li <shli@fusionio.com>
Signed-off-by: NeilBrown <neilb@suse.de>
Track overwrite disk count, so we can know if a stripe is a full stripe write.
Signed-off-by: Shaohua Li <shli@fusionio.com>
Signed-off-by: NeilBrown <neilb@suse.de>
A freshly new stripe with write request can be batched. Any time the stripe is
handled or new read is queued, the flag will be cleared.
Signed-off-by: Shaohua Li <shli@fusionio.com>
Signed-off-by: NeilBrown <neilb@suse.de>
Ilya Dryomov