| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Currently, Type_conversion_expression::do_is_constant thinks the
int-to-string conversion is constant if the integer operand is
constant, but Type_conversion_expression::do_get_backend actually
generates a call to runtime.intstring if the integer does not fit
in a "ushort", which makes it not suitable in constant context,
such as static initializer.
This CL makes it handle all constant integer input as constant,
generating constant string.
Fixes golang/go#32347.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/179777
From-SVN: r271821
|
|
Currently, goroutine switches are implemented with libc
getcontext/setcontext functions, which saves/restores the machine
register states and also the signal context. This does more than
what we need, and performs an expensive syscall.
This CL implements a simplified version of getcontext/setcontext,
in assembly, that only saves/restores the necessary part, i.e.
the callee-save registers, and the PC, SP. A simplified version
of makecontext, written in C, is also added. Currently this is
only implemented on Linux/AMD64.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/178298
From-SVN: r271818
|
|
gcc/go/
* go-gcc.cc (Gcc_backend::Gcc_backend): Add BUILT_IN_ATOMIC_FETCH_AND_1
and BUILT_IN_ATOMIC_FETCH_OR_1.
From-SVN: r271790
|
|
Let the Go frontend recognize sync/atomic functions and turn them
into intrinsics.
Also make sure not to intrinsify calls in go or defer statements.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/178937
From-SVN: r271784
|
|
loops which are not C++98)
PR go/90669
compiler: remove range-based 'for' loop
Fix for GCC PR/90669: remove range-based 'for' loop to preserve
buildability with g++ version 4.X.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/179397
From-SVN: r271761
|
|
PR go/90635
libgo: correct typo in USE_LIBFFI AM_CONDITIONAL
Only affects the case of passing --without-libffi to configure.
Fixes https://gcc.gnu.org/PR90635
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/178998
From-SVN: r271640
|
|
parameter ‘w’ [-Werror=unused-parameter] Continued (uint32_t *w))
PR go/90614
syscall: avoid unused parameter error if WIFCONTINUED not defined
Fixes https://gcc.gnu.org/PR90614
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/178997
From-SVN: r271638
|
|
The current scheme used by the compiler for "gcbits" symbols involves
generating a symbol name based on a 32-char encoding of the bits data.
This scheme works well in most cases but can generate very long symbol
names in rare cases. To help avoid such long symbol names, switch to a
different encoding scheme based on the SHA1 digest of the payload if
the symbol size would be too large.
Fixes golang/go#32083.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/177598
From-SVN: r271322
|
|
PR go/90482
compiler: make value method of direct interface type takes pointer
Currently, a value method of a direct interface type takes the
value of the receiver, which is pointer shaped, as the first
parameter. When this method is called through interface, we
actually pass the interface data as a pointer. On most platforms
this is ok, as the underlying calling convention is the same,
except that on SPARC32, the calling convention is actually
different.
This CL changes the method function actually takes a pointer.
The function will convert the pointer to the pointer-shaped
receiver type (a no-op conversion from machine code's aspect).
For a direct call, in the caller we convert the receiver to a
pointer (also no-op conversion) before invoking the method. For
an interface call, we pass the pointer as before. This way, it is
consistent that we always pass a pointer.
Hopefully this fixes SPARC32 build and https://gcc.gnu.org/PR90482.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/177758
From-SVN: r271310
|
|
Currently runtime/internal/atomic functions are implemented in C
using C compiler intrinsics. This CL lets the Go frontend
recognize these functions and turn them into intrinsics directly.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/176918
* go-gcc.cc (Gcc_backend::Gcc_backend): Define atomic builtins.
From-SVN: r271308
|
|
runtime/internal/sys.Ctz32/64 and Bswap32/64 are currently
implemented with compiler builtin functions. But if they are
called from another package, the compiler does not know and
therefore cannot turn them into compiler intrinsics. This CL
makes the compiler recognize these functions and turn them into
intrinsics directly, as the gc compiler does.
This CL sets up a way for adding intrinsics in the compiler.
More intrinsics will be added in later CLs.
Also move the handling of runtime.getcallerpc/sp to the new way
of generating intrinsics.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/176917
From-SVN: r271303
|
|
If an interface does not escape, it doesn't need a heap
allocation to hold the data (for non-direct interface type).
This CL improves the escape analysis to track interface
conversions, and reduces these allocations.
Implicit interface conversions were mostly added late in the
compilation pipeline, after the escape analysis. For the escape
analysis to see them, we move the introduction of these
conversions earlier, right before the escape analysis.
Now that the compiler can generate interface conversions inlined,
gcc/testsuite/go.test/test/nilptr2.go needs to be adjusted as in
golang.org/cl/176579, so the use function does an actual use.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/176459
* go.test/test/nilptr2.go: Change use function to actually do
something.
From-SVN: r271276
|
|
Revise the gccgo version of memory/block/mutex profiling to reduce
runtime overhead. The main change is to collect raw stack traces while
the profile is on line, then post-process the stacks just prior to the
point where we are ready to use the final product. Memory profiling
(at a very low sampling rate) is enabled by default, and the overhead
of the symbolization / DWARF-reading from backtrace_full was slowing
things down relative to the main Go runtime.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/171497
From-SVN: r271172
|
|
Based on patch by Rainer Orth.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/176938
From-SVN: r271135
|
|
runtime.throw needs a g to work properly. Set up g early, to
ensure that if something goes wrong in the runtime startup (e.g.
runtime.check fails), the program terminates in a reasonable way.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/176657
From-SVN: r271088
|
|
This does not permit any new inlinable functions in the standard library.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/176637
From-SVN: r271074
|
|
This adds all of two inlinable functions to the standard library:
crypto/subtle.ConstantTimeLessOrEq, regexp.(*Regexp).Copy.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/176378
From-SVN: r271063
|
|
If a string([]byte) conversion is used immediately as a key for a
map read, we don't need to copy the backing store of the byte
slice, as mapaccess does not keep a reference to it.
The gc compiler does more than this: it also avoids the copy if
the map key is a composite literal that contains the conversion
as a field, like, T{ ... { ..., string(b), ... }, ... }. For now,
we just optimize the simple case, which is probably most common.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/176197
* go.dg/mapstring.go: New test.
From-SVN: r271044
|
|
If a string([]byte) conversion is used immediately in a string
comparison, we don't need to copy the backing store of the byte
slice, as the string comparison doesn't hold any reference to
it. Instead, just create a string header from the byte slice and
pass it for comparison.
A new type of expression, String_value_expression, is introduced,
for constructing string headers.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/170894
* go.dg/cmpstring.go: New test.
From-SVN: r271021
|
|
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/176001
From-SVN: r271019
|
|
The builtin copy function is lowered to runtime functions
slicecopy, stringslicecopy, or typedslicecopy. The first two are
basically thin wrappers of memmove. Instead of making a runtime
call, we can just use __builtin_memmove. This gives the compiler
backend opportunities for further optimizations.
Move the lowering of builtin copy function to flatten phase for
the ease of rewriting.
Also do this optimization for the copy part of append(s1, s2...).
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/170005
From-SVN: r271017
|
|
We can use the intrinsic memmove directly, without going through
C.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/170004
* go-gcc.cc (Gcc_backend::Gcc_backend): Define memmove builtin.
From-SVN: r271016
|
|
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/175557
From-SVN: r271014
|
|
A direct interface type's value method takes value receiver now.
Don't pass pointer to the method function.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/175798
From-SVN: r271000
|
|
For a direct interface type T with a value method M, its pointer
type (*T)'s method table includes a stub method of M which takes
a (*T) as the receiver instead of a T. However, for the "typ"
field of the method table entry, we added another layer of
indirection, which makes it appear to take a **T, which is wrong.
This causes problems when using reflect.Type.Method to get the
method. This CL fixes the second, incorrect, indirection.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/175837
From-SVN: r270999
|
|
Add a -fgo-debug-optimization option to emit optimization
diagnostics. This can be used for testing optimizations. Apply
this to the range clear optimizations of maps and arrays.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/170002
gcc/go:
* lang.opt (-fgo-debug-optimization): New option.
* go-c.h (struct go_create_gogo_args): Add debug_optimization
field.
* go-lang.c (go_langhook_init): Set debug_optimization field.
* gccgo.texi (Invoking gccgo): Document -fgo-debug-optimization.
gcc/testsuite:
* go.dg/arrayclear.go: New test.
* go.dg/mapclear.go: New test.
From-SVN: r270993
|
|
Recognize
for i := range a { a[i] = zero }
for array or slice a, and rewrite it to call memclr, as the gc
compiler does.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/169398
From-SVN: r270862
|
|
The corresponding Go Toolchain patch is CL 164039
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/175079
From-SVN: r270857
|
|
Recognize
for k := range m { delete(m, k) }
for map m, and rewrite it to runtime.mapclear, as the gc compiler
does.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/169397
From-SVN: r270780
|
|
A direct interface is an interface whose data word contains the
actual data value, instead of a pointer to it. The gc toolchain
creates a direct interface if the value is pointer shaped, that
includes pointers (including unsafe.Pointer), functions, channels,
maps, and structs and arrays containing a single pointer-shaped
field. In gccgo, we only do this for pointers. This CL unifies
direct interface types with gc. This reduces allocations when
converting such types to interfaces.
Our method functions used to always take pointer receivers, to
make interface calls easy. Now for direct interface types, their
value methods will take value receivers. For a pointer to those
types, when converted to interface, the interface data contains
the pointer. For that interface to call a value method, it will
need a wrapper method that dereference the pointer and invokes
the value method. The wrapper method, instead of the actual one,
is put into the itable of the pointer type.
In the runtime, adjust funcPC for the new layout of interfaces of
functions.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/168409
From-SVN: r270779
|
|
Previously, each time we do an interface conversion for which the
method table is not known at compile time, we allocate a new
method table.
This CL ports the mechanism of itab caching from the gc runtime,
adapted to our itab representation and method finding mechanism.
With the cache, we reuse the same itab for the same (interface,
concrete) type pair. This reduces allocations in interface
conversions.
Unlike the gc runtime, we don't prepopulate the cache with
statically allocated itabs, as currently we don't have a way to
find them. This means we don't deduplicate run-time allocated
itabs with compile-time allocated ones. But that is not too bad
-- it is just a cache anyway.
As now itabs are never freed, it is also possible to drop the
write barrier for writing the first word of an interface header.
I'll leave this optimization for the future.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/171617
From-SVN: r270778
|
|
This avoids a compiler crash on invalid code.
Fixes https://gcc.gnu.org/PR90272
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/174377
From-SVN: r270658
|
|
AIX doesn't allow to mmap an address range which is already mmap.
Therefore, once the region has been allocated, it must munmap before
being able to play with it.
The corresponding Go Toolchain patch is CL 174059.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/174138
From-SVN: r270615
|
|
Fix a problem with Function_declaration::import_function relating to
how no-name or "sink" parameters are handled. In Gogo::start_function
(for the non-inline case) when parameter bindings are being added,
parameters with empty/sink names are renamed to synthesized "r.%d" /
"p.%d" names so as to avoid collisions. This same handling needs to be
present when creating the bindings for an inline function that's being
instantiated after being read from export data.
Fixes golang/go#31637.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/173538
From-SVN: r270564
|
|
PR target/89093
runtime: mark unwind functions general-regs-only on ARM
For https://gcc.gnu.org/PR89093.
Change-Id: Ic426b43d633c77104bda01d4e7835bc9ab4695ef
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/173657
Reviewed-by: Ian Lance Taylor <iant@golang.org>
From-SVN: r270542
|
|
This patch is required in order to build golang.org/x/net. The
corresponding Go Toolchain patch is CL 170537.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/172898
From-SVN: r270458
|
|
PR go/90110
compiler: use temporary to avoid early destruction
The code was passing a substr directly to strtol, and then checking
the *end value returned by strtol. But the substr could be destroyed
as soon as strtol returns, making the test of *end invalid.
Also fix an incorrect test of the string index rather than the value.
Fixes https://gcc.gnu.org/PR90110
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/172663
From-SVN: r270434
|
|
Resolve a small problem with concatenation of string constants: in a
string concat X + Y where X has named type and Y has abstract string
type, insure that the result has X's type, and disable folding if the
both sides have a concrete type that does not match.
Fixes golang/go#31412.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/171797
From-SVN: r270336
|
|
We can have multiple packages with the same name, so also sort by pkgpath.
To avoid an inconsistent sort, sort by symbol and pointer address if
we somehow get two different packages with the same name and pkgpath.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/171032
From-SVN: r270220
|
|
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/170706
From-SVN: r270214
|
|
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/168408
From-SVN: r269841
|
|
In the C calling convention, on AMD64, and probably a number of
other architectures, a 3-word struct argument is passed on stack.
This is less efficient than passing in three registers. Further,
this may affect the code generation in other part of the program,
even if the function is not actually called.
Slices are common in Go and append is a common slice operation,
which calls growslice in the growing path. To improve the code
generation, pass the slice header's three fields as separate
values, instead of a struct, to growslice.
The drawback is that this makes the runtime implementation
slightly diverges from the gc runtime.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/168277
From-SVN: r269811
|
|
Since aix/ppc64 has been added to GC toolchain, a mix between new and
old files were created in gcc toolchain.
This commit corrects this merge for aix/ppc64 and aix/ppc.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/167658
From-SVN: r269797
|
|
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/167749
From-SVN: r269780
|
|
When an inline function (with body) is imported from another package,
make that the "nointerface" property (if set) is preserved.
Fixes golang/go#30862.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/167742
From-SVN: r269713
|
|
Fixes bootstrap with a C++98 compiler.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/167762
From-SVN: r269710
|
|
The compiler generates two array index expressions when lowering
an append expression. Currently they generate bound checks.
Bound checks are not necessary in this case, as we know the slice
has, or will grow to, enough length and capacity. Eliminate them.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/166817
From-SVN: r269699
|
|
This change fixes a bug in which two interface types were being
incorrectly commoned (considered identical) in the initial stages of
writing out types to export data. The indexer does a walk to collect
candidates for export, inserting types into a table to eliminate
duplicates; as part of this process a local interface type T1 was
being commoned with a different interface type T2. This caused a cycle
in the exported type graph due to the way embedded interfaces are
handled.
The fix was to add a new flag to the Type::is_identical utility
routine to request that interface type comparison be done by examining
the original parse methods, as opposed to the expanded method set,
then use the new flag when creating the hash map for the exporter.
Fixes golang/go#30659.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/166638
From-SVN: r269634
|
|
Add a new debugging utility routine debug_go_type(), intended to
display the contents of a Type object in a way useful to debugging
a run of the compiler. Prior to this the only useful alternative
for debugging types was invoking the mangled_name() method, which
has problems (for example, won't work on interface types prior
to finalizing of methods).
This is a "deep" dump, meaning that all types reachable from the
type passed to debug_go_type() will be printed out. Example:
(gdb) print debug_go_type(t1)
T0 0x535f300 'net/http.Header' -> T1
T1 0x535d3d0 map ['string' -> string] T4
T2 0x5304bb0 'string' -> string
T3 0x331f900 string
T4 0x535d370 array [] 'string' -> string
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/166637
From-SVN: r269633
|
|
PR go/89447
syscall, internal/syscall: adjust use of largefile functions
Consistently call __go_openat for openat. Use fstatat64, creat64,
sendfile64, and getdents64 where needed.
Based on patch by Rainer Orth.
Fixes https://gcc.gnu.org/PR89447
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/166420
From-SVN: r269521
|
