On 6l and 8l, this is a real instruction, guaranteed to
cause an 'undefined instruction' exception.
On 5l, we simulate it as BL to address 0.
The plan is to use it as a signal to the linker that this
point in the instruction stream cannot be reached
(hence the changes to nofollow). This will help the
compiler explain that panicindex and friends do not
return without having to put a list of these functions
in the linker.
R=ken2
CC=golang-dev
https://golang.org/cl/6255064
16 seems pretty standard on x86 for function entry.
I don't know if ARM would benefit, so I used just 4
(single instruction alignment).
This has a minor absolute effect on the current timings.
The main hope is that it will make them more consistent from
run to run.
benchmark old ns/op new ns/op delta
BenchmarkBinaryTree17 4222117400 4140739800 -1.93%
BenchmarkFannkuch11 3462631800 3259914400 -5.85%
BenchmarkGobDecode 20887622 20620222 -1.28%
BenchmarkGobEncode 9548772 9384886 -1.72%
BenchmarkGzip 151687 150333 -0.89%
BenchmarkGunzip 8742 8741 -0.01%
BenchmarkJSONEncode 62730560 65210990 +3.95%
BenchmarkJSONDecode 252569180 249394860 -1.26%
BenchmarkMandelbrot200 5267599 5273394 +0.11%
BenchmarkRevcomp25M 980813500 996013800 +1.55%
BenchmarkTemplate 361259100 360620840 -0.18%
R=ken2
CC=golang-dev
https://golang.org/cl/6244066
The code was inconsistent about when it used
brchain(x) and when it used x directly, with the result
that you could end up emitting code for brchain(x) but
leave the jump pointing at an unemitted x.
R=ken2
CC=golang-dev
https://golang.org/cl/6250077
The old code generated for a bounds check was
CMP
JLT ok
CALL panicindex
ok:
...
The new code is (once the linker finishes with it):
CMP
JGE panic
...
panic:
CALL panicindex
which moves the calls out of line, putting more useful
code in each cache line. This matters especially in tight
loops, such as in Fannkuch. The benefit is more modest
elsewhere, but real.
From test/bench/go1, amd64:
benchmark old ns/op new ns/op delta
BenchmarkBinaryTree17 6096092000 6088808000 -0.12%
BenchmarkFannkuch11 6151404000 4020463000 -34.64%
BenchmarkGobDecode 28990050 28894630 -0.33%
BenchmarkGobEncode 12406310 12136730 -2.17%
BenchmarkGzip 179923 179903 -0.01%
BenchmarkGunzip 11219 11130 -0.79%
BenchmarkJSONEncode 86429350 86515900 +0.10%
BenchmarkJSONDecode 334593800 315728400 -5.64%
BenchmarkRevcomp25M 1219763000 1180767000 -3.20%
BenchmarkTemplate 492947600 483646800 -1.89%
And 386:
benchmark old ns/op new ns/op delta
BenchmarkBinaryTree17 6354902000 6243000000 -1.76%
BenchmarkFannkuch11 8043769000 7326965000 -8.91%
BenchmarkGobDecode 19010800 18941230 -0.37%
BenchmarkGobEncode 14077500 13792460 -2.02%
BenchmarkGzip 194087 193619 -0.24%
BenchmarkGunzip 12495 12457 -0.30%
BenchmarkJSONEncode 125636400 125451400 -0.15%
BenchmarkJSONDecode 696648600 685032800 -1.67%
BenchmarkRevcomp25M 2058088000 2052545000 -0.27%
BenchmarkTemplate 602140000 589876800 -2.04%
To implement this, two new instruction forms:
JLT target // same as always
JLT $0, target // branch expected not taken
JLT $1, target // branch expected taken
The linker could also emit the prediction prefixes, but it
does not: expected taken branches are reversed so that the
expected case is not taken (as in example above), and
the default expectaton for such a jump is not taken
already.
R=golang-dev, gri, r, dave
CC=golang-dev
https://golang.org/cl/6248049
Introduce a newsym() to cmd/lib.c to add a symbol but don't add
them to hash table.
Introduce a new bit flag SHIDDEN and bit mask SMASK to handle hidden
and/or local symbols in ELF symbol tables. Though we still need to order
the symbol table entries correctly.
Fix for issue 3261 comment #9.
For CL 5822049.
R=iant, rsc
CC=golang-dev
https://golang.org/cl/5823055
Some newer Linux distributions (Ubuntu ARM at least) use a new multiarch
directory organization, where dynamic linker is no longer in the hardcoded
path in our linker.
For example, Ubuntu 12.04 ARM hardfloat places its dynamic linker at
/lib/arm-linux-gnueabihf/ld-linux.so.3
Ref: http://lackof.org/taggart/hacking/multiarch/
Also, to support Debian GNU/kFreeBSD as a FreeBSD variant, we need this capability, so it's part of issue 3533.
This CL add a new pragma (#pragma dynlinker "path") to cc.
R=iant, rsc
CC=golang-dev
https://golang.org/cl/6086043
dodata will convert to SNOPTRDATA if appropriate.
Should fix arm build (hope springs eternal).
TBR=golang-dev
CC=golang-dev
https://golang.org/cl/5687074
cc: add #pragma textflag to set it
runtime: mark mheap to go into noptr-bss.
remove special case in garbage collector
Remove the ARM from.flag field created by CL 5687044.
The DUPOK flag was already in p->reg, so keep using that.
Otherwise test/nilptr.go creates a very large binary.
Should fix the arm build.
Diagnosed by minux.ma; replacement for CL 5690044.
R=golang-dev, minux.ma, r
CC=golang-dev
https://golang.org/cl/5686060
The garbage collector can avoid scanning this section, with
reduces collection time as well as the number of false positives.
Helps a little bit with issue 909, but certainly does not solve it.
R=ken2
CC=golang-dev
https://golang.org/cl/5671099
8g/cgen.c
print format type mismatch
8l/asm.c
resoff set and not used
gc/pgen.c
misleading comparison INT > 0x80000000
gc/reflect.c
dalgsym must be static to match forward declaration
gc/subr.c
assumed_equal set and not used
hashmem's second argument is not used
gc/walk.c
duplicated (unreachable) code
R=rsc
CC=golang-dev
https://golang.org/cl/5651079
As a convenience to people working on the tools,
leave Makefiles that invoke the go dist tool appropriately.
They are not used during the build.
R=golang-dev, bradfitz, n13m3y3r, gustavo
CC=golang-dev
https://golang.org/cl/5636050
Also delete gotest, since it's messy to fix and slated for deletion anyway.
A couple of things outside src can't be tested any more. "go test" will be
fixed and these tests will be re-enabled. They're noisy for now.
Fixes#284.
R=rsc
CC=golang-dev
https://golang.org/cl/5598049
To allow these types as map keys, we must fill in
equal and hash functions in their algorithm tables.
Structs or arrays that are "just memory", like [2]int,
can and do continue to use the AMEM algorithm.
Structs or arrays that contain special values like
strings or interface values use generated functions
for both equal and hash.
The runtime helper func runtime.equal(t, x, y) bool handles
the general equality case for x == y and calls out to
the equal implementation in the algorithm table.
For short values (<= 4 struct fields or array elements),
the sequence of elementwise comparisons is inlined
instead of calling runtime.equal.
R=ken, mpimenov
CC=golang-dev
https://golang.org/cl/5451105
If the length of the interpreter string
pushes us over the ELFRESERVE limit, the
resulting error message will be comical.
I was doing some ELF tinkering with a
modified version of 8l when I hit this.
To be clear, the stock linkers wouldn't
hit this without adding about forty more
section headers. We're safe for now. ;)
Also, remove a redundant call to cflush.
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/5268044
This requires making the .dynamic section writable, as the
dynamic linker will change the value of the DT_DEBUG tag at
runtime. The DT_DEBUG tag is used by gdb to find all loaded
shared libraries.
R=rsc
CC=golang-dev
https://golang.org/cl/5189044
This line was triggering a null dereference warning
under clang-3.0. The line was added in a46819aa9150
but compared to it's sibling in 6l it appears to be
leftover debugging.
R=rsc
CC=golang-dev
https://golang.org/cl/5049042
The dynamic ELF sections were pointing to the proper data,
but that data was already owned by the rodata and text sections.
Some ELF references explicitly prohibit multiple sections from
owning the same data, and strip behaves accordingly.
The data for these sections was moved out and their ranges are
now owned by their respective sections. This change makes strip
happy both with and without -s being provided at link time.
A test was added in debug/elf to ensure there are no regressions
on this area in the future.
Fixes#1242.
Fixes#2022.
NOTE: Tested on Linux amd64/386/arm only.
R=rsc
CC=golang-dev
https://golang.org/cl/4808043
Reduces number of write+seek's from 88516 to 2080
when linking godoc with 6l.
Thanks to Alex Brainman for pointing out the
many small writes.
R=golang-dev, r, alex.brainman, robert.hencke
CC=golang-dev
https://golang.org/cl/4743043
Per the TIS ELF spec, if a PHDR entry is present in the
program header table, it must be part of the memory image of
the program. Failure to do this makes elflint complain, and
causes some tools that manipulate ELF to crash.
R=iant, rsc
CC=dave, golang-dev
https://golang.org/cl/4650067
The gosymtab and gopclntab sections were pointing to the proper
data, but that data was already owned by the rodata section.
Some ELF references explicitly prohibit multiple sections from
owning the same data, and strip behaves accordingly.
The data for these sections was moved to after rodata, and the
gosymtab and gopclntab sections now own their respective ranges.
This change makes strip happy both with and without -s being
provided at link time. Note that it won't remove these sections
because they are still allocated, and that's by design since
they are necessary at runtime for generating proper backtraces
and similar introspection operations.
Unlike the previous behavior, -s will now maintain zero-sized
gosymtab and gopclntab sections. This makes the implementation
slightly cleaner.
Fixes#1242.
NOTE: Tested on Linux amd64/386/arm only.
R=ality, rsc
CC=golang-dev
https://golang.org/cl/4639077
Russ Cox
Shenghou Ma
Jan Ziak
Evan Shaw
Darren Elwood
Anthony Martin
Rob Pike
Wei Guangjing
Joel Sing
Christopher Nielsen
Ian Lance Taylor
Lucio De Re
Mikkel Krautz
Dave Cheney
Yuval Pavel Zholkover
Jaroslavas Počepko
Dmitriy Vyukov
Gustavo Niemeyer
David Anderson