Now all the fixes are in place, let's rock-n-roll!
Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Anything that uses a constructed instruction (ie. from ppc-opcode.h),
need to use the new R0 macro, as %r0 is not going to work.
Also convert usages of macros where we are just determining an offset
(usually for a load/store), like:
std r14,STK_REG(r14)(r1)
Can't use STK_REG(r14) as %r14 doesn't work in the STK_REG macro since
it's just calculating an offset.
Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The assembler doesn't take %r0 register arguments in braces, so remove them.
Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
We are going to use these later and convert r0 to %r0 etc.
Signed-off-by: Michael Neuling <mikey@neuling.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Newer gcc are being a bit blind here (it's pretty obvious we don't
reach the code path using the array if we haven't initialized the
pointer) but none of that is performance critical so let's just
silence it.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
There was a typo, checking for CONFIG_TRACE_IRQFLAG instead of
CONFIG_TRACE_IRQFLAGS causing some useful debug code to not be
built
This in turns causes a build error on BookE 64-bit due to incorrect
semicolons at the end of a couple of macros, so let's fix that too
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: stable@vger.kernel.org [v3.4]
Looks like we still have issues with pSeries and Cell idle code
vs. the lazy irq state. In fact, the reset fixes that went upstream
are exposing the problem more by causing BUG_ON() to trigger (which
this patch turns into a WARN_ON instead).
We need to be careful when using a variant of low power state that
has the side effect of turning interrupts back on, to properly set
all the SW & lazy state to look as if everything is enabled before
we enter the low power state with MSR:EE off as we will return with
MSR:EE on. If not, we have a discrepancy of state which can cause
things to go very wrong later on.
This patch moves the logic into a helper and uses it from the
pseries and cell idle code. The power4/970 idle code already got
things right (in assembly even !) so I'm not touching it. The power7
"bare metal" idle code is subtly different and correct. Remains PA6T
and some hypervisor based Cell platforms which have questionable
code in there, but they are mostly dead platforms so I'll fix them
when I manage to get final answers from the respective maintainers
about how the low power state actually works on them.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
CC: stable@vger.kernel.org [v3.4]
Pull ARM fixes from Russell King:
"Last merge window, we had some updates from Al cleaning up the signal
restart handling. These have caused some problems on ARM, and while
Al has some fixes, we have some concerns with Al's patches but we've
been unsuccesful with discussing this.
We have got to the point where we need to do something, and we've
decided that the best solution is to revert the appropriate commits
until Al is able to reply to us.
Also included here are four patches to fix warnings that I've noticed
in my build system, and one fix for kprobes test code."
* 'fixes' of git://git.linaro.org/people/rmk/linux-arm:
ARM: fix warning caused by wrongly typed arm_dma_limit
ARM: fix warnings about atomic64_read
ARM: 7440/1: kprobes: only test 'sub pc, pc, #1b-2b+8-2' on ARMv6
ARM: 7441/1: perf: return -EOPNOTSUPP if requested mode exclusion is unavailable
ARM: 7443/1: Revert "new way of handling ERESTART_RESTARTBLOCK"
ARM: 7442/1: Revert "remove unused restart trampoline"
ARM: fix set_domain() macro
ARM: fix mach-versatile/pci.c warning
Small fixes on multiple ARM platforms
* A build regression from a previous fix on dove and mv78xx0
* Two fixes for recently (3.5-rc1) changed mmp/pxa code
* multiple omap2+ bug fixes
* two trivial fixes for i.MX
* one v3.5 regression for mxs
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)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=2Ynd
-----END PGP SIGNATURE-----
Merge tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Pull ARM SoC fixes from Arnd Bergmann:
"Small fixes on multiple ARM platforms
- A build regression from a previous fix on dove and mv78xx0
- Two fixes for recently (3.5-rc1) changed mmp/pxa code
- multiple omap2+ bug fixes
- two trivial fixes for i.MX
- one v3.5 regression for mxs"
* tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc:
ARM: apx4devkit: fix FEC enabling PHY clock
ARM: OMAP2+: hwmod data: Fix wrong McBSP clock alias on OMAP4
ARM: OMAP4: hwmod data: temporarily comment out data for the usb_host_fs and aess IP blocks
ARM: Orion: Fix WDT compile for Dove and MV78xx0
ARM: mmp: remove mach/gpio-pxa.h
ARM: imx: assert SCC gate stays enabled
ARM: OMAP4: TWL6030: ensure sys_nirq1 is mux'd and wakeup enabled
ARM: OMAP2: Overo: init I2C before MMC to fix MMC suspend/resume failure
ARM: imx27_visstrim_m10: Do not include <asm/system.h>
ARM: pxa: hx4700: Fix basic suspend/resume
Pull KVM fix from Marcelo Tosatti:
"Memory leak and oops on the x86 mmu code, and sanitization of the
KVM_IRQFD ioctl."
* git://git.kernel.org/pub/scm/virt/kvm/kvm:
KVM: MMU: fix shrinking page from the empty mmu
KVM: fix fault page leak
KVM: Sanitize KVM_IRQFD flags
KVM: Add missing KVM_IRQFD API documentation
KVM: Pass kvm_irqfd to functions
arch/arm/mm/init.c: In function 'arm_memblock_init':
arch/arm/mm/init.c:380: warning: comparison of distinct pointer types lacks a cast
by fixing the typecast in its definition when DMA_ZONE is disabled.
This was missed in 4986e5c7c (ARM: mm: fix type of the arm_dma_limit
global variable).
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Fix:
net/netfilter/xt_connbytes.c: In function 'connbytes_mt':
net/netfilter/xt_connbytes.c:43: warning: passing argument 1 of 'atomic64_read' discards qualifiers from pointer target type
...
by adding the missing const.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
'sub pc, pc, #1b-2b+8-2' results in address<1:0> == '10'.
sub pc, pc, #const (== ADR pc, #const) performs an interworking branch
(BXWritePC()) on ARMv7+ and a simple branch (BranchWritePC()) on earlier
versions.
In ARM state, BXWritePC() is UNPREDICTABLE when address<1:0> == '10'.
In ARM state on ARMv6+, BranchWritePC() ignores address<1:0>. Before
ARMv6, BranchWritePC() is UNPREDICTABLE if address<1:0> != '00'
So the instruction is UNPREDICTABLE both before and after v6.
Acked-by: Jon Medhurst <tixy@yxit.co.uk>
Signed-off-by: Rabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
working again.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)
iQIcBAABAgAGBQJP9VYqAAoJEBvUPslcq6VzwCIP/1/0vqJgr4rSayTXnhxGaRkb
rPCFORJySrY7de1bVuVFpo3IfRV1jTDTcCYNSIRaV3Ph1OIKM/W+WFddzRRJ4OFP
E4NmBC8bPg875SES0lUkX6KmsLCR5W+LYFh2j+2qrESmrE95MRLgnuYXyezNoF1m
2L27J+PbkIZpjXq+jXRZT6wzYk3+cPwCV7oNYWoH7YKHSAqa6h3ywW+7+urMdwap
W9tnmb3mwgTGSQv5ThrP1BS3l2xFkTVW3xyMSeIeFGvp3gK/tAAgMhLhkCVjGilX
amOCmEX7SupZNwWuIEnbnDxxhlr5gSowGN4xDetkppZ10K18fVWia6y61Ls6AE7V
8ZXL7/YaJ5OTkznWs1NmVDZ7dJRtlIMD4MFzYdMXWthl9af30vojpgZs7/Ikiu2G
1xEUTYMZzyb77IvbbR3oeMF1WRfmcW/B0b16MFpgqdV4Awaj63IHHwIzhWgL41XP
1xIr9RZtnjE1859UJIETanUc4L+0fqpl6brUIO5pNoS9VEbCyADzBicmZxk0JRmx
JugrNt7Z53qlWbd17umg0oiWCmAOK2DTI9ZHzn/A/w+ksQPvGwiGUzFpPniZ+g5t
3jz/CXQ5Au4OWb23EFNrL5/2kqxSU6hTxxtQJF+5MDuiHltsAl8/IGQQs5X6zl3v
TnSl8nZVvWzFLxotsLW8
=9LmJ
-----END PGP SIGNATURE-----
Merge tag 'omap-fixes-for-v3.5-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap into fixes
PM related fixes for omaps mostly to get suspend/resume
working again.
* tag 'omap-fixes-for-v3.5-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap:
ARM: OMAP2+: hwmod data: Fix wrong McBSP clock alias on OMAP4
ARM: OMAP4: hwmod data: temporarily comment out data for the usb_host_fs and aess IP blocks
ARM: OMAP4: TWL6030: ensure sys_nirq1 is mux'd and wakeup enabled
ARM: OMAP2: Overo: init I2C before MMC to fix MMC suspend/resume failure
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
We currently return -EPERM if the user requests mode exclusion that is
not supported by the CPU. This looks pretty confusing from userspace
and is inconsistent with other architectures (ppc, x86).
This patch returns -EOPNOTSUPP instead.
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
This reverts commit 6b5c8045ec.
Conflicts:
arch/arm/kernel/ptrace.c
The new syscall restarting code can lead to problems if we take an
interrupt in userspace just before restarting the svc instruction. If
a signal is delivered when returning from the interrupt, the
TIF_SYSCALL_RESTARTSYS will remain set and cause any syscalls executed
from the signal handler to be treated as a restart of the previously
interrupted system call. This includes the final sigreturn call, meaning
that we may fail to exit from the signal context. Furthermore, if a
system call made from the signal handler requires a restart via the
restart_block, it is possible to clear the thread flag and fail to
restart the originally interrupted system call.
The right solution to this problem is to perform the restarting in the
kernel, avoiding the possibility of handling a further signal before the
restart is complete. Since we're almost at -rc6, let's revert the new
method for now and aim for in-kernel restarting at a later date.
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
This reverts commit fa18484d09.
We need the restart trampoline back so that we can revert a related
problematic patch 6b5c8045ec ("arm: new
way of handling ERESTART_RESTARTBLOCK").
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Avoid polluting drivers with a set_domain() macro, which interferes with
structure member names:
drivers/net/wireless/ath/ath9k/dfs_pattern_detector.c:294:33: error: macro "set_domain" passed 2 arguments, but takes just 1
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
management and McBSP.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)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=GhIQ
-----END PGP SIGNATURE-----
Merge tag 'omap-fixes-b-for-3.5rc' of git://git.kernel.org/pub/scm/linux/kernel/git/pjw/omap-pending into fixes
A few more OMAP fixes for 3.5-rc. These fix some bugs with power
management and McBSP.
Ethernet stopped to work after mxs clk framework change.
Signed-off-by: Lauri Hintsala <lauri.hintsala@bluegiga.com>
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
arch/arm/mach-versatile/pci.c: In function 'versatile_map_irq':
arch/arm/mach-versatile/pci.c:342: warning: unused variable 'devslot'
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
The commit 503d0ea24d
ARM: OMAP4: hwmod data: Add aliases for McBSP fclk clocks
added a wrong "prcm_clk" alias for PRCM clock whereas the McBSP
driver and previous OMAPs are using "prcm_fck".
It thus lead to the following warning.
[ 47.409729] omap-mcbsp: clks: could not clk_get() prcm_fck
Fix that by changing the opt_clk role to prcm_fck.
Reported-by: Misael Lopez Cruz <misael.lopez@ti.com>
Signed-off-by: Benoit Cousson <b-cousson@ti.com>
Cc: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Sebastien Guiriec <s-guiriec@ti.com>
Signed-off-by: Paul Walmsley <paul@pwsan.com>
The OMAP4 usb_host_fs (OHCI) and AESS IP blocks require some special
programming for them to enter idle. Without this programming, they
will prevent the rest of the chip from entering full chip idle.
To implement the idle programming cleanly, this will take some
coordination between maintainers. This is likely to take some time,
so it is probably best to leave this for 3.6 or 3.7. So, in the
meantime, prevent these IP blocks from being registered.
Later, once the appropriate support is available, this patch can be
reverted.
This second version comments out the IP block data since Benoît didn't
like removing it.
Signed-off-by: Paul Walmsley <paul@pwsan.com>
Cc: Benoît Cousson <b-cousson@ti.com>
Commit 0fa1f0609a (ARM: Orion: Fix
Virtual/Physical mixup with watchdog) broke the Dove & MV78xx0
build. Although these two SoC don't use the watchdog, the shared
platform code still needs to build. Add the necessary defines.
Cc: stable@vger.kernel.org
Reported-by: Nicolas Pitre <nico@fluxnic.net>
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Tested-by: Nicolas Pitre <nico@fluxnic.net>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Commit 157d2644cb ("ARM: pxa: change gpio
to platform device") removed all includes of mach/gpio-pxa.h. It kept
this unused header in the tree. Using it can't work, as it itself
includes the non-existent header plat/gpio-pxa.h. This header can safely
be removed.
Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
Acked-by: Eric Miao <eric.y.miao@gmail.com>
Signed-off-by: Haojian Zhuang <haojian.zhuang@gmail.com>
The SCC clock is needed in internal boot mode and so must keep enabled.
This same issue was fixed for the pre-common-clk code in commit
3d6e614 (mx35: Fix boot ROM hang in internal boot mode)
Cc: John Ogness <jogness@linutronix.de>
Cc: Hans J. Koch <hjk@hansjkoch.de>
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
The array of names in hugetlbpage.c no longer exists.
Signed-off-by: Scott Wood <scottwood@freescale.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Commit 430b01e8f5 ("[POWERPC] Kill
flatdevtree.c") killed the two files including flatdevtree_env.h. It was
apparently just an oversight to not kill that header too. Kill it now.
Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Warning(arch/powerpc/kernel/pci_of_scan.c:210): Excess function parameter 'node' description in 'of_scan_pci_bridge'
Warning(arch/powerpc/kernel/vio.c:636): No description found for parameter 'desired'
Warning(arch/powerpc/kernel/vio.c:636): Excess function parameter 'new_desired' description in 'vio_cmo_set_dev_desired'
Warning(arch/powerpc/kernel/vio.c:1270): No description found for parameter 'viodrv'
Warning(arch/powerpc/kernel/vio.c:1270): Excess function parameter 'drv' description in '__vio_register_driver'
Warning(arch/powerpc/kernel/vio.c:1289): No description found for parameter 'viodrv'
Warning(arch/powerpc/kernel/vio.c:1289): Excess function parameter 'driver' description in 'vio_unregister_driver'
Signed-off-by: Wanpeng Li <liwp@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
memblock_end_of_DRAM() returns end_address + 1, not end address.
While some code assumes that it returns end address.
Signed-off-by: Bharat Bhushan <bharat.bhushan@freescale.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
I blame Mikey for this. He elevated my slightly dubious testcase:
to benchmark status. And naturally we need to be number 1 at creating
zeros. So lets improve __clear_user some more.
As Paul suggests we can use dcbz for large lengths. This patch gets
the destination cacheline aligned then uses dcbz on whole cachelines.
Before:
10485760000 bytes (10 GB) copied, 0.414744 s, 25.3 GB/s
After:
10485760000 bytes (10 GB) copied, 0.268597 s, 39.0 GB/s
39 GB/s, a new record.
Signed-off-by: Anton Blanchard <anton@samba.org>
Tested-by: Olof Johansson <olof@lixom.net>
Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
At the moment all queues in a multiqueue adapter will serialise
against the IOMMU table lock. This is proving to be a big issue,
especially with 10Gbit ethernet.
This patch creates 4 pools and tries to spread the load across
them. If the table is under 1GB in size we revert back to the
original behaviour of 1 pool and 1 largealloc pool.
We create a hash to map CPUs to pools. Since we prefer interrupts to
be affinitised to primary CPUs, without some form of hashing we are
very likely to end up using the same pool. As an example, POWER7
has 4 way SMT and with 4 pools all primary threads will map to the
same pool.
The largealloc pool is reduced from 1/2 to 1/4 of the space to
partially offset the overhead of breaking the table up into pools.
Some performance numbers were obtained with a Chelsio T3 adapter on
two POWER7 boxes, running a 100 session TCP round robin test.
Performance improved 69% with this patch applied.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
In preparation for IOMMU pools, push the spinlock into
iommu_range_alloc and __iommu_free.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
This patch moves tce_free outside of the lock in iommu_free.
Some performance numbers were obtained with a Chelsio T3 adapter on
two POWER7 boxes, running a 100 session TCP round robin test.
Performance improved 25% with this patch applied.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
We currently hold the IOMMU spinlock around tce_build and tce_flush.
This causes our spinlock hold times to be much higher than required
and can impact multiqueue adapters.
This patch moves tce_build and tce_flush outside of the lock in
iommu_alloc, and tce_flush outside of the lock in iommu_free.
Some performance numbers were obtained with a Chelsio T3 adapter on
two POWER7 boxes, running a 100 session TCP round robin test.
Performance improved 32% with this patch applied.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
tce_buildmulti_pSeriesLP uses a per cpu page to communicate with the
hypervisor. We currently rely on the IOMMU table spinlock but
subsequent patches will be removing that so disable interrupts
around all accesses of tce_page.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Implement a POWER7 optimised memcpy using VMX and enhanced prefetch
instructions.
This is a copy of the POWER7 optimised copy_to_user/copy_from_user
loop. Detailed implementation and performance details can be found in
commit a66086b819 (powerpc: POWER7 optimised
copy_to_user/copy_from_user using VMX).
I noticed memcpy issues when profiling a RAID6 workload:
.memcpy
.async_memcpy
.async_copy_data
.__raid_run_ops
.handle_stripe
.raid5d
.md_thread
I created a simplified testcase by building a RAID6 array with 4 1GB
ramdisks (booting with brd.rd_size=1048576):
# mdadm -CR -e 1.2 /dev/md0 --level=6 -n4 /dev/ram[0-3]
I then timed how long it took to write to the entire array:
# dd if=/dev/zero of=/dev/md0 bs=1M
Before: 892 MB/s
After: 999 MB/s
A 12% improvement.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Version 2.06 of the POWER ISA introduced enhanced touch instructions,
allowing us to specify a number of attributes including the length of
a stream.
This patch adds a software stream for both loads and stores in the
POWER7 copy_tofrom_user loop. Since the setup is quite complicated
and we have to use an eieio to ensure correct ordering of the "GO"
command we only do this for copies above 4kB.
To quantify any performance improvements we need a working set
bigger than the caches so we operate on a 1GB file:
# dd if=/dev/zero of=/tmp/foo bs=1M count=1024
And we compare how fast we can read the file:
# dd if=/tmp/foo of=/dev/null bs=1M
before: 7.7 GB/s
after: 9.6 GB/s
A 25% improvement.
The worst case for this patch will be a completely L1 cache contained
copy of just over 4kB. We can test this with the copy_to_user
testcase we used to tune copy_tofrom_user originally:
http://ozlabs.org/~anton/junkcode/copy_to_user.c
# time ./copy_to_user2 -l 4224 -i 10000000
before: 6.807 s
after: 6.946 s
A 2% slowdown, which seems reasonable considering our data is unlikely
to be completely L1 contained.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
While creating the PCI root bus through function pci_create_root_bus()
of PCI core, it should have assigned the secondary bus number for the
newly created PCI root bus. Thus we needn't do the explicit assignment
for the secondary bus number again in pcibios_scan_phb().
Signed-off-by: Gavin Shan <shangw@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The form affinity for NUMA is set to 1 if the firmware supports
OPAL. Otherwise, we have to retrieve that from OF node "/chosen".
For the latter case, OF node "/chosen" reference count was never
decreased.
Signed-off-by: Gavin Shan <shangw@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Implement a POWER7 optimised copy_page using VMX and enhanced
prefetch instructions. We use enhanced prefetch hints to prefetch
both the load and store side. We copy a cacheline at a time and
fall back to regular loads and stores if we are unable to use VMX
(eg we are in an interrupt).
The following microbenchmark was used to assess the impact of
the patch:
http://ozlabs.org/~anton/junkcode/page_fault_file.c
We test MAP_PRIVATE page faults across a 1GB file, 100 times:
# time ./page_fault_file -p -l 1G -i 100
Before: 22.25s
After: 18.89s
17% faster
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Subsequent patches will add more VMX library functions and it makes
sense to keep all the c-code helper functions in the one file.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
mtmsrd is an expensive instruction, we save a few cycles by
doing it once instead of twice.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Version 2.06 of the POWER ISA introduced enhanced touch instructions,
allowing us to specify a number of attributes including the length of
a stream.
This patch adds a software stream for both loads and stores in the
POWER7 copy_tofrom_user loop. Since the setup is quite complicated
and we have to use an eieio to ensure correct ordering of the "GO"
command we only do this for copies above 4kB.
To quantify any performance improvements we need a working set
bigger than the caches so we operate on a 1GB file:
# dd if=/dev/zero of=/tmp/foo bs=1M count=1024
And we compare how fast we can read the file:
# dd if=/tmp/foo of=/dev/null bs=1M
before: 7.7 GB/s
after: 9.6 GB/s
A 25% improvement.
The worst case for this patch will be a completely L1 cache contained
copy of just over 4kB. We can test this with the copy_to_user
testcase we used to tune copy_tofrom_user originally:
http://ozlabs.org/~anton/junkcode/copy_to_user.c
# time ./copy_to_user2 -l 4224 -i 10000000
before: 6.807 s
after: 6.946 s
A 2% slowdown, which seems reasonable considering our data is unlikely
to be completely L1 contained.
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Michael Neuling