| Age | Commit message (Collapse) | Author | Lines |
|---|
|
The shared get_futex_key code does not work on nommu, but is not
needed anyway because it's impossible for a given backing to have
multiple distinct virtual addresses on nommu. Simply disable these
code paths by refraining from setting FLAG_SHARED when CONFIG_MMU is
not enabled.
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
This is a hack to fix a nasty bug: the rescan is completely
unsynchronized with use of the spi bus for spi-based mmc, and thus
clobbers in-progress transfers when it happens. No idea what the right
fix is.
|
|
|
|
This is sloppy and not ready for upstream.
|
|
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
The SH2 version of entry.S uses global variables, which need to be
cpu-local in order to work with SMP. For ease of access from asm,
simply use arrays indexed by cpu number, and require the availability
of an address (mmio register or properly setup per-cpu memory) from
which the current cpu's index can be read.
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
The version of futex.h in asm-generic should really be adapted to do
the same thing so that this hideous code does not have to be
duplicated per-arch.
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
The J-Core "spi2" device is a PIO-based SPI master controller. It
differs from "bitbang" devices in that that it's clocked in hardware
rather than via soft clock modulation over gpio, and performs
byte-at-a-time transfers between the cpu and SPI controller.
This driver will be extended to support future versions of the J-Core
SPI controller with DMA transfers when they become available.
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
This defconfig is intended not to be specific to a particular board;
it enables drivers for all currently-supported hardware, and should be
updated to include additional drivers as they are added.
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
At the hardware level, the J-Core PIT is integrated with the interrupt
controller, but it is represented as its own device and has an
independent programming interface. It provides a 12-bit countdown
timer, which is not presently used, and a periodic timer. The interval
length for the latter is programmable via a 32-bit throttle register
whose units are determined by a bus-period register. The periodic
timer is used to implement both periodic and oneshot clock event
modes; in oneshot mode the interrupt handler simply disables the timer
as soon as it fires.
Despite its device tree node representing an interrupt for the PIT,
the actual irq generated is programmable, not hard-wired. The driver
is responsible for programming the PIT to generate the hardware irq
number that the DT assigns to it.
On SMP configurations, J-Core provides cpu-local instances of the PIT;
no broadcast timer is needed. This driver supports the creation of the
necessary per-cpu clock_event_device instances. The code has been
tested and works on SMP, but will not be usable without additional
J-Core SMP-support patches and appropriate hardware capable of running
SMP.
A nanosecond-resolution clocksource is provided using the J-Core "RTC"
registers, which give a 64-bit seconds count and 32-bit nanoseconds.
The driver converts these to a 64-bit nanoseconds count.
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
There are two versions of the J-Core interrupt controller in use, aic1
which generates interrupts with programmable priorities, but only
supports 8 irq lines and maps them to cpu traps in the range 17 to 24,
and aic2 which uses traps in the range 64-127 and supports up to 128
irqs, with priorities dependent on the interrupt number. The Linux
driver does not make use of priorities anyway.
For simplicity, there is no aic1-specific logic in the driver beyond
setting the priority register, which is necessary for interrupts to
work at all. Eventually aic1 will likely be phased out, but it's
currently in use in deployments and all released bitstream binaries.
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
At the CPU/ISA level, the J2 is compatible with SH-2, and thus the
changes to add J2 support build on existing SH-2 support. However, J2
does not duplicate the memory-mapped SH-2 features like the cache
interface. Instead, the cache interfaces is described in the device
tree, and new code is added to be able to access the flat device tree
at early boot before it is unflattened.
Support is also added for receiving interrupts on trap numbers in the
range 16 to 31, since the J-Core aic1 interrupt controller generates
these traps. This range was unused but nominally for hardware
exceptions on SH-2, and a few values in this range were used for
exceptions on SH-2A, but SH-2A has its own version of the relevant
code.
No individual cpu subtypes are added for J2 since the intent moving
forward is to represent SoCs with device tree rather than as
hard-coded subtypes in the kernel. The CPU_SUBTYPE_J2 Kconfig item
exists only to fit into the existing cpu selection mechanism until it
is overhauled.
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
Signed-off-by: Rich Felker <dalias@libc.org>
Acked-by: Rob Herring <robh@kernel.org>
|
|
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
The J-Core project (j-core.org) produces open source cpu and SoC
peripheral cores synthesizable as FPGA bitstreams or ASICs.
Signed-off-by: Rich Felker <dalias@libc.org>
Acked-by: Rob Herring <robh@kernel.org>
|
|
Such a configuration could only be selected by manually selecting
CONFIG_OF; SH_DEVICE_TREE selects both. The affected code is using the
flat DTB at boot time and thus rightfully should depend on
OF_FLATTREE, not just OF.
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
There is no arch-specific sched_clock implementation for sh, resulting
in use of the old default jiffies-based implementation. Instead, use
the modern generic sched_clock framework so that drivers can register
better backends.
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
The Kconfig for this driver is currently:
config HEARTBEAT
bool "Heartbeat LED"
....meaning that it currently is not being built as a module by anyone.
Lets remove the modular code that is essentially orphaned, so that
when reading the driver there is no doubt it is builtin-only.
Since module_init translates to device_initcall in the non-modular
case, the init ordering remains unchanged with this commit.
We explicitly disallow a driver unbind, since that doesn't have a
sensible use case anyway, and it allows us to drop the ".remove"
code for non-modular drivers.
We also delete the MODULE_LICENSE tag etc. since all that information
is already contained at the top of the file in the comments.
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Rich Felker <dalias@libc.org>
Cc: linux-sh@vger.kernel.org
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
The Kconfig currently controlling compilation of this code is:
config SH_SECUREEDGE5410
bool "SecureEdge5410"
....meaning that it currently is not being built as a module by anyone.
Lets remove the couple traces of modularity so that when reading the
driver there is no doubt it is builtin-only.
Since module_init translates to device_initcall in the non-modular
case, the init ordering remains unchanged with this commit.
We don't replace module.h with init.h since the file already has that.
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Rich Felker <dalias@libc.org>
Cc: linux-sh@vger.kernel.org
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
The Makefile/Kconfig currently controlling compilation of this code is:
obj-$(CONFIG_DEBUG_FS) += $(debugfs-y)
debugfs-y := asids-debugfs.o
lib/Kconfig.debug:config DEBUG_FS
lib/Kconfig.debug: bool "Debug Filesystem"
....meaning that it currently is not being built as a module by anyone.
Lets remove the couple traces of modular code, so that when reading the
driver there is no doubt it is builtin-only.
Since module_init translates to device_initcall in the non-modular
case, the init ordering remains unchanged with this commit.
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Rich Felker <dalias@libc.org>
Cc: linux-sh@vger.kernel.org
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
The Makefile currently controlling compilation of this code is:
obj-y := debugtraps.o dma-nommu.o dumpstack.o \
[...]
syscalls_$(BITS).o time.o topology.o traps.o \
traps_$(BITS).o unwinder.o
....meaning that it currently is not being built as a module by anyone.
Lets remove the couple traces of modular code, so that when reading
the driver there is no doubt it is builtin-only.
Since module_init translates to device_initcall in the non-modular
case, the init ordering remains unchanged with this commit.
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Rich Felker <dalias@libc.org>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Cc: linux-sh@vger.kernel.org
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
The futex cmpxchg runtime testing in kernel/futex.c depends on
accesses to address 0 producing EFAULT, which obviously does not work
on nommu. Since SH always has cmpxchg, disable the broken runtime
detection.
At some point this should be fixed at the kernel/futex.c level. UP
machines can always provide a working cmpxchg with interrupt masking,
and SMP cannot function without a working cmpxchg anyway.
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
SH3/4 (with MMU) have a virtually indexed cache, requiring explicit
work to avoid consistency problems arising from having the same
physical address range cached in multiple cache lines. This is
unneeded for the NOMMU case, and some of the resulting code paths
(kmap_coherent) don't work. SH2 only avoided this problem by having a
4-way associative cache with way size equal to the page size (4k),
yielding no cache index bits outside of the page offset and thus no
aliases.
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
Up until now, the SH version of the sigcontext structure, and thus
mcontext_t/ucontext_t, varied depending on the cpu model the kernel
was built to run on. SH-4 (including SH-4A) and SH-2A used the form
with space for FPU registers, and everything else used a form that
omitted them.
From a userspace perspective, however, the structure layout must be
fixed for a given ABI. Traditionally glibc and uClibc used the form
with space for FPU registers only when __SH4__ (which implies FPU;
__SH4_NOFPU__ is the predefined macro for SH-4 but with no-FPU ABI)
was defined. As a result:
- SH-4 no-FPU programs never matched kernel sigcontext.
- SH-3 programs did not match kernel sigcontext if run on SH-4,
despite an apparent intent that they be compatible.
- SH-2 and SH-2A programs (using uClibc) did not match kernel
sigcontext if run on SH-2A.
The mismatch might seem inconsequential because it occurs at the end
of the sigcontext structure, but sigcontext is embedded as uc_mcontext
in ucontext_t, where it is followed by uc_sigmask, an important member
for signal handlers to have access to. In particular, access to
uc_sigmask is necessary for a correct implementation of thread
cancellation.
It would be possible to retain support for both sigcontext ABIs via a
personality mechanism, but since many configurations were already
broken and nobody noticed, and since there are very few if any users
of legacy no-FPU models anymore, I have opted to just remove the
variation and always include space for the FPU registers in
sigcontext. This was proposed and discussed on a thread "SH sigcontext
ABI is broken" cross-posted to linux-sh, libc-alpha, and musl libc
lists in June 2015, and no objections were raised.
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
Correct bitoff in big endian OS.
Current code works correctly for 1 byte but not for 2 bytes.
Fixes: 3226aad81aa6 ("sh: support 1 and 2 byte xchg")
Signed-off-by: Pan Xinhui <xinhui.pan@linux.vnet.ibm.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Rich Felker <dalias@libc.org>
|
|
|
|
The self-test was updated to cover zero-length strings; the function
needs to be updated, too.
Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: George Spelvin <linux@sciencehorizons.net>
Fixes: fcfd2fbf22d2 ("fs/namei.c: Add hashlen_string() function")
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
The original name was simply hash_string(), but that conflicted with a
function with that name in drivers/base/power/trace.c, and I decided
that calling it "hashlen_" was better anyway.
But you have to do it in two places.
[ This caused build errors for architectures that don't define
CONFIG_DCACHE_WORD_ACCESS - Linus ]
Signed-off-by: George Spelvin <linux@sciencehorizons.net>
Reported-by: Guenter Roeck <linux@roeck-us.net>
Fixes: fcfd2fbf22d2 ("fs/namei.c: Add hashlen_string() function")
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
The HPFS filesystem used generic_show_options to produce string that is
displayed in /proc/mounts. However, there is a problem that the options
may disappear after remount. If we mount the filesystem with option1
and then remount it with option2, /proc/mounts should show both option1
and option2, however it only shows option2 because the whole option
string is replaced with replace_mount_options in hpfs_remount_fs.
To fix this bug, implement the hpfs_show_options function that prints
options that are currently selected.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
Commit c8f33d0bec99 ("affs: kstrdup() memory handling") checks if the
kstrdup function returns NULL due to out-of-memory condition.
However, if we are remounting a filesystem with no change to
filesystem-specific options, the parameter data is NULL. In this case,
kstrdup returns NULL (because it was passed NULL parameter), although no
out of memory condition exists. The mount syscall then fails with
ENOMEM.
This patch fixes the bug. We fail with ENOMEM only if data is non-NULL.
The patch also changes the call to replace_mount_options - if we didn't
pass any filesystem-specific options, we don't call
replace_mount_options (thus we don't erase existing reported options).
Fixes: c8f33d0bec99 ("affs: kstrdup() memory handling")
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: stable@vger.kernel.org # v4.1+
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
Commit ce657611baf9 ("hpfs: kstrdup() out of memory handling") checks if
the kstrdup function returns NULL due to out-of-memory condition.
However, if we are remounting a filesystem with no change to
filesystem-specific options, the parameter data is NULL. In this case,
kstrdup returns NULL (because it was passed NULL parameter), although no
out of memory condition exists. The mount syscall then fails with
ENOMEM.
This patch fixes the bug. We fail with ENOMEM only if data is non-NULL.
The patch also changes the call to replace_mount_options - if we didn't
pass any filesystem-specific options, we don't call
replace_mount_options (thus we don't erase existing reported options).
Fixes: ce657611baf9 ("hpfs: kstrdup() out of memory handling")
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
Pull more MIPS updates from Ralf Baechle:
"This is the secondnd batch of MIPS patches for 4.7. Summary:
CPS:
- Copy EVA configuration when starting secondary VPs.
EIC:
- Clear Status IPL.
Lasat:
- Fix a few off by one bugs.
lib:
- Mark intrinsics notrace. Not only are the intrinsics
uninteresting, it would cause infinite recursion.
MAINTAINERS:
- Add file patterns for MIPS BRCM device tree bindings.
- Add file patterns for mips device tree bindings.
MT7628:
- Fix MT7628 pinmux typos.
- wled_an pinmux gpio.
- EPHY LEDs pinmux support.
Pistachio:
- Enable KASLR
VDSO:
- Build microMIPS VDSO for microMIPS kernels.
- Fix aliasing warning by building with `-fno-strict-aliasing' for
debugging but also tracing them might result in recursion.
Misc:
- Add missing FROZEN hotplug notifier transitions.
- Fix clk binding example for varioius PIC32 devices.
- Fix cpu interrupt controller node-names in the DT files.
- Fix XPA CPU feature separation.
- Fix write_gc0_* macros when writing zero.
- Add inline asm encoding helpers.
- Add missing VZ accessor microMIPS encodings.
- Fix little endian microMIPS MSA encodings.
- Add 64-bit HTW fields and fix its configuration.
- Fix sigreturn via VDSO on microMIPS kernel.
- Lots of typo fixes.
- Add definitions of SegCtl registers and use them"
* 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus: (49 commits)
MIPS: Add missing FROZEN hotplug notifier transitions
MIPS: Build microMIPS VDSO for microMIPS kernels
MIPS: Fix sigreturn via VDSO on microMIPS kernel
MIPS: devicetree: fix cpu interrupt controller node-names
MIPS: VDSO: Build with `-fno-strict-aliasing'
MIPS: Pistachio: Enable KASLR
MIPS: lib: Mark intrinsics notrace
MIPS: Fix 64-bit HTW configuration
MIPS: Add 64-bit HTW fields
MAINTAINERS: Add file patterns for mips device tree bindings
MAINTAINERS: Add file patterns for mips brcm device tree bindings
MIPS: Simplify DSP instruction encoding macros
MIPS: Add missing tlbinvf/XPA microMIPS encodings
MIPS: Fix little endian microMIPS MSA encodings
MIPS: Add missing VZ accessor microMIPS encodings
MIPS: Add inline asm encoding helpers
MIPS: Spelling fix lets -> let's
MIPS: VR41xx: Fix typo
MIPS: oprofile: Fix typo
MIPS: math-emu: Fix typo
...
|
|
Various builds (such as i386:allmodconfig) fail with
fs/binfmt_aout.c:133:2: error: expected identifier or '(' before 'return'
fs/binfmt_aout.c:134:1: error: expected identifier or '(' before '}' token
[ Oops. My bad, I had stupidly thought that "allmodconfig" covered this
on x86-64 too, but it obviously doesn't. Egg on my face. - Linus ]
Fixes: 5d22fc25d4fc ("mm: remove more IS_ERR_VALUE abuses")
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
|
|
Pull string hash improvements from George Spelvin:
"This series does several related things:
- Makes the dcache hash (fs/namei.c) useful for general kernel use.
(Thanks to Bruce for noticing the zero-length corner case)
- Converts the string hashes in <linux/sunrpc/svcauth.h> to use the
above.
- Avoids 64-bit multiplies in hash_64() on 32-bit platforms. Two
32-bit multiplies will do well enough.
- Rids the world of the bad hash multipliers in hash_32.
This finishes the job started in commit 689de1d6ca95 ("Minimal
fix-up of bad hashing behavior of hash_64()")
The vast majority of Linux architectures have hardware support for
32x32-bit multiply and so derive no benefit from "simplified"
multipliers.
The few processors that do not (68000, h8/300 and some models of
Microblaze) have arch-specific implementations added. Those
patches are last in the series.
- Overhauls the dcache hash mixing.
The patch in commit 0fed3ac866ea ("namei: Improve hash mixing if
CONFIG_DCACHE_WORD_ACCESS") was an off-the-cuff suggestion.
Replaced with a much more careful design that's simultaneously
faster and better. (My own invention, as there was noting suitable
in the literature I could find. Comments welcome!)
- Modify the hash_name() loop to skip the initial HASH_MIX(). This
would let us salt the hash if we ever wanted to.
- Sort out partial_name_hash().
The hash function is declared as using a long state, even though
it's truncated to 32 bits at the end and the extra internal state
contributes nothing to the result. And some callers do odd things:
- fs/hfs/string.c only allocates 32 bits of state
- fs/hfsplus/unicode.c uses it to hash 16-bit unicode symbols not bytes
- Modify bytemask_from_count to handle inputs of 1..sizeof(long)
rather than 0..sizeof(long)-1. This would simplify users other
than full_name_hash"
Special thanks to Bruce Fields for testing and finding bugs in v1. (I
learned some humbling lessons about "obviously correct" code.)
On the arch-specific front, the m68k assembly has been tested in a
standalone test harness, I've been in contact with the Microblaze
maintainers who mostly don't care, as the hardware multiplier is never
omitted in real-world applications, and I haven't heard anything from
the H8/300 world"
* 'hash' of git://ftp.sciencehorizons.net/linux:
h8300: Add <asm/hash.h>
microblaze: Add <asm/hash.h>
m68k: Add <asm/hash.h>
<linux/hash.h>: Add support for architecture-specific functions
fs/namei.c: Improve dcache hash function
Eliminate bad hash multipliers from hash_32() and hash_64()
Change hash_64() return value to 32 bits
<linux/sunrpc/svcauth.h>: Define hash_str() in terms of hashlen_string()
fs/namei.c: Add hashlen_string() function
Pull out string hash to <linux/stringhash.h>
|
|
This will improve the performance of hash_32() and hash_64(), but due
to complete lack of multi-bit shift instructions on H8, performance will
still be bad in surrounding code.
Designing H8-specific hash algorithms to work around that is a separate
project. (But if the maintainers would like to get in touch...)
Signed-off-by: George Spelvin <linux@sciencehorizons.net>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: uclinux-h8-devel@lists.sourceforge.jp
|
|
Microblaze is an FPGA soft core that can be configured various ways.
If it is configured without a multiplier, the standard __hash_32()
will require a call to __mulsi3, which is a slow software loop.
Instead, use a shift-and-add sequence for the constant multiply.
GCC knows how to do this, but it's not as clever as some.
Signed-off-by: George Spelvin <linux@sciencehorizons.net>
Cc: Alistair Francis <alistair.francis@xilinx.com>
Cc: Michal Simek <michal.simek@xilinx.com>
|
|
This provides a multiply by constant GOLDEN_RATIO_32 = 0x61C88647
for the original mc68000, which lacks a 32x32-bit multiply instruction.
Yes, the amount of optimization effort put in is excessive. :-)
Shift-add chain found by Yevgen Voronenko's Hcub algorithm at
http://spiral.ece.cmu.edu/mcm/gen.html
Signed-off-by: George Spelvin <linux@sciencehorizons.net>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Greg Ungerer <gerg@linux-m68k.org>
Cc: Andreas Schwab <schwab@linux-m68k.org>
Cc: Philippe De Muyter <phdm@macq.eu>
Cc: linux-m68k@lists.linux-m68k.org
|
|
This is just the infrastructure; there are no users yet.
This is modelled on CONFIG_ARCH_RANDOM; a CONFIG_ symbol declares
the existence of <asm/hash.h>.
That file may define its own versions of various functions, and define
HAVE_* symbols (no CONFIG_ prefix!) to suppress the generic ones.
Included is a self-test (in lib/test_hash.c) that verifies the basics.
It is NOT in general required that the arch-specific functions compute
the same thing as the generic, but if a HAVE_* symbol is defined with
the value 1, then equality is tested.
Signed-off-by: George Spelvin <linux@sciencehorizons.net>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Greg Ungerer <gerg@linux-m68k.org>
Cc: Andreas Schwab <schwab@linux-m68k.org>
Cc: Philippe De Muyter <phdm@macq.eu>
Cc: linux-m68k@lists.linux-m68k.org
Cc: Alistair Francis <alistai@xilinx.com>
Cc: Michal Simek <michal.simek@xilinx.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: uclinux-h8-devel@lists.sourceforge.jp
|
|
Patch 0fed3ac866 improved the hash mixing, but the function is slower
than necessary; there's a 7-instruction dependency chain (10 on x86)
each loop iteration.
Word-at-a-time access is a very tight loop (which is good, because
link_path_walk() is one of the hottest code paths in the entire kernel),
and the hash mixing function must not have a longer latency to avoid
slowing it down.
There do not appear to be any published fast hash functions that:
1) Operate on the input a word at a time, and
2) Don't need to know the length of the input beforehand, and
3) Have a single iterated mixing function, not needing conditional
branches or unrolling to distinguish different loop iterations.
One of the algorithms which comes closest is Yann Collet's xxHash, but
that's two dependent multiplies per word, which is too much.
The key insights in this design are:
1) Barring expensive ops like multiplies, to diffuse one input bit
across 64 bits of hash state takes at least log2(64) = 6 sequentially
dependent instructions. That is more cycles than we'd like.
2) An operation like "hash ^= hash << 13" requires a second temporary
register anyway, and on a 2-operand machine like x86, it's three
instructions.
3) A better use of a second register is to hold a two-word hash state.
With careful design, no temporaries are needed at all, so it doesn't
increase register pressure. And this gets rid of register copying
on 2-operand machines, so the code is smaller and faster.
4) Using two words of state weakens the requirement for one-round mixing;
we now have two rounds of mixing before cancellation is possible.
5) A two-word hash state also allows operations on both halves to be
done in parallel, so on a superscalar processor we get more mixing
in fewer cycles.
I ended up using a mixing function inspired by the ChaCha and Speck
round functions. It is 6 simple instructions and 3 cycles per iteration
(assuming multiply by 9 can be done by an "lea" instruction):
x ^= *input++;
y ^= x; x = ROL(x, K1);
x += y; y = ROL(y, K2);
y *= 9;
Not only is this reversible, two consecutive rounds are reversible:
if you are given the initial and final states, but not the intermediate
state, it is possible to compute both input words. This means that at
least 3 words of input are required to create a collision.
(It also has the property, used by hash_name() to avoid a branch, that
it hashes all-zero to all-zero.)
The rotate constants K1 and K2 were found by experiment. The search took
a sample of random initial states (I used 1023) and considered the effect
of flipping each of the 64 input bits on each of the 128 output bits two
rounds later. Each of the 8192 pairs can be considered a biased coin, and
adding up the Shannon entropy of all of them produces a score.
The best-scoring shifts also did well in other tests (flipping bits in y,
trying 3 or 4 rounds of mixing, flipping all 64*63/2 pairs of input bits),
so the choice was made with the additional constraint that the sum of the
shifts is odd and not too close to the word size.
The final state is then folded into a 32-bit hash value by a less carefully
optimized multiply-based scheme. This also has to be fast, as pathname
components tend to be short (the most common case is one iteration!), but
there's some room for latency, as there is a fair bit of intervening logic
before the hash value is used for anything.
(Performance verified with "bonnie++ -s 0 -n 1536:-2" on tmpfs. I need
a better benchmark; the numbers seem to show a slight dip in performance
between 4.6.0 and this patch, but they're too noisy to quote.)
Special thanks to Bruce fields for diligent testing which uncovered a
nasty fencepost error in an earlier version of this patch.
[checkpatch.pl formatting complaints noted and respectfully disagreed with.]
Signed-off-by: George Spelvin <linux@sciencehorizons.net>
Tested-by: J. Bruce Fields <bfields@redhat.com>
|
|
The "simplified" prime multipliers made very bad hash functions, so get rid
of them. This completes the work of 689de1d6ca.
To avoid the inefficiency which was the motivation for the "simplified"
multipliers, hash_64() on 32-bit systems is changed to use a different
algorithm. It makes two calls to hash_32() instead.
drivers/media/usb/dvb-usb-v2/af9015.c uses the old GOLDEN_RATIO_PRIME_32
for some horrible reason, so it inherits a copy of the old definition.
Signed-off-by: George Spelvin <linux@sciencehorizons.net>
Cc: Antti Palosaari <crope@iki.fi>
Cc: Mauro Carvalho Chehab <m.chehab@samsung.com>
|
|
That's all that's ever asked for, and it makes the return
type of hash_long() consistent.
It also allows (upcoming patch) an optimized implementation
of hash_64 on 32-bit machines.
I tried adding a BUILD_BUG_ON to ensure the number of bits requested
was never more than 32 (most callers use a compile-time constant), but
adding <linux/bug.h> to <linux/hash.h> breaks the tools/perf compiler
unless tools/perf/MANIFEST is updated, and understanding that code base
well enough to update it is too much trouble. I did the rest of an
allyesconfig build with such a check, and nothing tripped.
Signed-off-by: George Spelvin <linux@sciencehorizons.net>
|
|
Finally, the first use of previous two patches: eliminate the
separate ad-hoc string hash functions in the sunrpc code.
Now hash_str() is a wrapper around hash_string(), and hash_mem() is
likewise a wrapper around full_name_hash().
Note that sunrpc code *does* call hash_mem() with a zero length, which
is why the previous patch needed to handle that in full_name_hash().
(Thanks, Bruce, for finding that!)
This also eliminates the only caller of hash_long which asks for
more than 32 bits of output.
The comment about the quality of hashlen_string() and full_name_hash()
is jumping the gun by a few patches; they aren't very impressive now,
but will be improved greatly later in the series.
Signed-off-by: George Spelvin <linux@sciencehorizons.net>
Tested-by: J. Bruce Fields <bfields@redhat.com>
Acked-by: J. Bruce Fields <bfields@redhat.com>
Cc: Jeff Layton <jlayton@poochiereds.net>
Cc: linux-nfs@vger.kernel.org
|
|
We'd like to make more use of the highly-optimized dcache hash functions
throughout the kernel, rather than have every subsystem create its own,
and a function that hashes basic null-terminated strings is required
for that.
(The name is to emphasize that it returns both hash and length.)
It's actually useful in the dcache itself, specifically d_alloc_name().
Other uses in the next patch.
full_name_hash() is also tweaked to make it more generally useful:
1) Take a "char *" rather than "unsigned char *" argument, to
be consistent with hash_name().
2) Handle zero-length inputs. If we want more callers, we don't want
to make them worry about corner cases.
Signed-off-by: George Spelvin <linux@sciencehorizons.net>
|
