The sighander has been run at the bottom of the
currently executing goroutine stack, but it's in C,
and we don't want C on our ordinary goroutine stacks.
Worse, it does a lot of stuff, and it might need more
stack space. There is scary code in traceback_windows.go
that talks about stack splits during sighandler.
Moving sighandler to g0 will eliminate the possibility
of stack splits and such, and then we can delete
traceback_windows.go entirely. Win win.
On the builder, all.bat passes with GOARCH=amd64
and all.bat gets most of the way with GOARCH=386
except for a DLL-loading test that I think is unrelated.
Fixes windows build.
TBR=brainman, iant
CC=golang-codereviews
https://golang.org/cl/140380043
This CL contains compiler+runtime changes that detect C code
running on Go (not g0, not gsignal) stacks, and it contains
corrections for what it detected.
The detection works by changing the C prologue to use a different
stack guard word in the G than Go prologue does. On the g0 and
gsignal stacks, that stack guard word is set to the usual
stack guard value. But on ordinary Go stacks, that stack
guard word is set to ^0, which will make any stack split
check fail. The C prologue then calls morestackc instead
of morestack, and morestackc aborts the program with
a message about running C code on a Go stack.
This check catches all C code running on the Go stack
except NOSPLIT code. The NOSPLIT code is allowed,
so the check is complete. Since it is a dynamic check,
the code must execute to be caught. But unlike the static
checks we've been using in cmd/ld, the dynamic check
works with function pointers and other indirect calls.
For example it caught sigpanic being pushed onto Go
stacks in the signal handlers.
Fixes#8667.
LGTM=khr, iant
R=golang-codereviews, khr, iant
CC=golang-codereviews, r
https://golang.org/cl/133700043
avoid tight coupling between deferreturn and jmpdefer.
before, jmpdefer knew the exact frame size of deferreturn
in order to pop it off the stack. now, deferreturn passes
jmpdefer a pointer to the frame above it explicitly.
that avoids a magic constant and should be less fragile.
R=r
DELTA=32 (6 added, 3 deleted, 23 changed)
OCL=29801
CL=29804
The code was already careful not to use malloc/free
for stack growth during calls to malloc.
Avoid them during calls to free too.
R=r
DELTA=9 (7 added, 0 deleted, 2 changed)
OCL=29606
CL=29610
* string format changed
* files got renamed
* new files that i forgot to check in last time
updates are all copy and paste from 6g
R=ken
OCL=29385
CL=29400
this is not a user-visible change.
before, all interface values were
struct Itype {
Sigt *type;
Sigi *inter;
void *method[n];
}
struct Iface {
void *addr;
Itype *itype;
}
the itype is basically a vtable, but it's unnecessary
if the static type is interface{ }.
for interface values with static type empty, the
new representation is
struct Eface {
void *addr;
Sigt *type;
}
this complicates the code somewhat, but
it reduces the number of Itypes that
have to be computed and cached,
it opens up opportunities to avoid function
calls in a few common cases,
and it will make it possible to lay out
interface{} values at compile time,
which i think i'll need for the new reflection.
R=ken
OCL=28701
CL=29121
* byteastring is no longer used
* do not generate ODCL, OAS for globals
(wasn't generating any code but might
save one or two init functions)
* do not call self from Init function
R=ken
OCL=28309
CL=28309
Currently a http server will be killed on receipt of SIGPIPE
if a client closes a socket which the server is trying to
write to.
R=rsc
APPROVED=rsc
DELTA=2 (0 added, 0 deleted, 2 changed)
OCL=27959
CL=28055
add test for string range.
test has minor failure: after loop the index == len(s); should be len(s)-1
in this case. according to spec, vars are left at position at last
iteration.
R=ken,rsc
DELTA=259 (161 added, 96 deleted, 2 changed)
OCL=27343
CL=27343
160 - 75 was just barely not enough for deferproc + morestack.
added enum names and bumped to 256 - 128.
added explanation.
changed a few mal() (garbage-collected) to
malloc()/free() (manually collected).
R=ken
OCL=26981
CL=26981
remove use of _subv in vlrt.c
darwin/386/signal.c
darwin/386/*
linux/386/* (forgotten before)
can run empty program on darwin/386 now.
R=r
DELTA=1140 (1021 added, 114 deleted, 5 changed)
OCL=26942
CL=26968
move darwin-amd64 specific code into runtime/darwin/amd64/
repeat for linux.
move rt0 into runtime.a instead of keeping a separate .6 file.
6l seems to have no problem with that.
TBR=r
OCL=26680
CL=26680
trying to find all the places where
spans might be recorded.
Free can cascade into complicated
span manipulations that move them
from list to list; the old code had the
possibility of accidentally processing
a span twice or jumping to a different
list, causing an infinite loop.
R=r
DELTA=70 (28 added, 25 deleted, 17 changed)
OCL=23704
CL=23710
though they still build in src/runtime.
use cgo instead of hand-written wrappers.
R=r
DELTA=740 (289 added, 300 deleted, 151 changed)
OCL=23326
CL=23331
implement hiding lowercase methods m in
signatures by adding in a hash of the package name
to the type hash code.
remove remaining checks for internally-generated _ names:
they are all gone.
R=ken
OCL=23236
CL=23238
lets us use a single source file for both 6c and gcc, handling
the incompatible handling of return values.
R=rsc
DELTA=649 (613 added, 35 deleted, 1 changed)
OCL=22682
CL=22730
+ Use macros to name symbols with non-ASCII characters.
+ Make some variables unsigned, because they are compared
against unsigned values.
+ Fix a few void* pointers to be MLink*.
R=rsc
DELTA=94 (44 added, 3 deleted, 47 changed)
OCL=22303
CL=22638