Rather than change the value of R13 during the execution, keep R13 fixed
(after the initial prelude) and always use offsets from it.
This should help the runtime figure out what's going on if, say, a
signal should occur while running this code.
I've also trimmed the set of saved registers since Go doesn't require
the callee to maintain anything except R10 and R13.
Change-Id: Ifbeca73c1d964cc43bb7f8c20c61066f22fd562d
Reviewed-on: https://go-review.googlesource.com/31717
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>
5f31782cfb added build constraints to
disable assembly for Go 1.6 but didn't add the needed tags to the ARM
files. Also, it's not clear that was needed as the error given in
golang/go#17424 only complains about the chacha20poly1305 package.
This change reenables the assembly for Go 1.6 in the poly1305 package.
Tested with 1.6.3 and 1.5.4.
Fixesgolang/go#17512.
Change-Id: I81b41f8810437ea327b415542402cd8ff5c8a390
Reviewed-on: https://go-review.googlesource.com/31492
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Up till now, sum_arm.s was working only because of luck. It was written
assuming that it had stack space below the current stack pointer, but Go
decrements the stack pointer in the function prelude, so it was just
writing off the end of the stack.
This change fixes the stack manipulation so that it only writes within
the bounds.
Fixesgolang/go#17499.
Change-Id: I1951b3344c21f6bd6ade79da8b96dd1bb68180db
Reviewed-on: https://go-review.googlesource.com/31443
Reviewed-by: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
The ARM assembly doesn't work for NaCl on ARM because it doesn't meet
the required rules. This change disables it on ARM and also fixes the
issue that the build constraints in sum_arm.s would be ignored because
they came after the #include.
Change-Id: I6cb3815ec62ac4686a6e72f405af104293586bb6
Reviewed-on: https://go-review.googlesource.com/31264
Run-TryBot: Adam Langley <agl@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
This change updates the Poly1305 code in x/crypto to reflect some
comments from the review of
https://go-review.googlesource.com/cl/29245/.
Following this change, poly1305_arm.s will be renamed to sum_arm.s, to
match the other files here. (The review becomes confusing if that's done
in the same change as the asmfmt changes.)
Change-Id: Iddf43615eba97c975adb135aef3a814a37e9ec02
Reviewed-on: https://go-review.googlesource.com/30820
Run-TryBot: Adam Langley <agl@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Michael Munday <munday@ca.ibm.com>
The current ARM implementation assumes that the input message
is memory aligned and so it can cause alignment fault when it
is not enabled. Also it may generate incorrect outputs in ARMv5.
This change fixes this issue by temporarily copying the input
to a local aligned space. Although there may be a better way
to handle unaligned access, this would be a safe way in all
ARM versions.
This change also added a test and benchmarks with unaligned
data. The benchmark result on RasberryPI 2 is
Benchmark64 2000000 812 ns/op 78.81 MB/s
Benchmark1K 200000 7809 ns/op 131.12 MB/s
Benchmark64Unaligned 2000000 967 ns/op 66.13 MB/s
Benchmark1KUnaligned 200000 10316 ns/op 99.26 MB/s
Change-Id: I189cc1b7bb6c67a04c9877271fb27326f2896e82
Reviewed-on: https://go-review.googlesource.com/12797
Reviewed-by: Adam Langley <agl@golang.org>
Fix compilation of poly1305 using go tip - it currently fails with:
./poly1305_arm.s:124: cannot reference SP without a symbol
./poly1305_arm.s:161: cannot reference SP without a symbol
./poly1305_arm.s:162: cannot reference SP without a symbol
asm: asm: assembly of ./poly1305_arm.s failed
Change-Id: I797dcf3641cc881b6cc192034b693ccf58317987
Reviewed-on: https://go-review.googlesource.com/10307
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Reviewed-by: Adam Langley <agl@golang.org>
This change adds ARMv6 assembly implementation. The referenced assembly code was
the public domain source by Andrew Moon in https://github.com/floodyberry/poly1305-opt/blob/master/app/extensions/poly1305/poly1305_armv6-32.inc.
The author has confirmed that it's ok to put it under the Go license.
Benchmark results on Raspberry Pi (ARMv6-compatible processor rev 7),
o Without ARMv6 assembly
Benchmark1K 5000 287177 ns/op 3.57 MB/s
Benchmark64 50000 38880 ns/op 1.65 MB/s
o With ARMv6 assembly
Benchmark1K 100000 15964 ns/op 64.14 MB/s
Benchmark64 1000000 1472 ns/op 43.46 MB/s
Change-Id: Iea5b0b831ac097cc6d477a8fccbf0ddb4819724c
Reviewed-on: https://go-review.googlesource.com/9765
Reviewed-by: Adam Langley <agl@golang.org>
Run-TryBot: Adam Langley <agl@golang.org>
Fixes tests when using Go tip. Without this the link steps
fails with errors like:
missing Go type information for global symbol: google3/third_party/golang/go_crypto/curve25519/curve25519.REDMASK51 size 8
LGTM=agl
R=agl
CC=golang-codereviews
https://golang.org/cl/156810043
This change alters the assembly to use FSUBD instructions such that
6l will actually emit the correct FSUBRD instructions and enables
the assembly code.
R=golang-dev, r
CC=golang-dev
https://golang.org/cl/6514044
Shenghou Ma
Adam Langley
Brad Fitzpatrick
Andreas Auernhammer
Jungho Ahn
Alexander Neumann
Joel Sing
Marga Manterola
David Symonds
Ian Lance Taylor
Shenghou Ma