code review 100590044: build: make nacl pass

build: make nacl pass

Add nacl.bash, the NaCl version of all.bash.
It's a separate script because it builds a variant of package syscall
with a large zip file embedded in it, containing all the input files
needed for tests.

Disable various tests new since the last round, mostly the ones using os/exec.

Fixes issue 7945.

[rsc, 2014-05-20 05:39:34]

Hello golang-codereviews@googlegroups.com (cc: dfc, remyoudompheng),

I'd like you to review this change to
https://code.google.com/p/go/

[remyoudompheng, 2014-05-20 05:47:40]

https://codereview.appspot.com/100590044/diff/60001/src/nacl.bash
File src/nacl.bash (right):

https://codereview.appspot.com/100590044/diff/60001/src/nacl.bash#newcode36
src/nacl.bash:36: GOOS=$GOHOSTOS GOARCH=$GOHOSTARCH $gobin/go run
../misc/nacl/mkzip.go -p syscall -r .. ../misc/nacl/testzip.proto
pkg/syscall/fstest_nacl.go
I can't find misc/nacl/mkzip.go , is it in a separate review ?

[dave_cheney.net, 2014-05-20 05:54:09]

Thanks Russ. Testing now.

https://codereview.appspot.com/100590044/diff/60001/src/cmd/cgo/gcc.go
File src/cmd/cgo/gcc.go (right):

https://codereview.appspot.com/100590044/diff/60001/src/cmd/cgo/gcc.go#newcod...
src/cmd/cgo/gcc.go:1203: if dt.ByteSize < 0 && tag == "" { // opaque unnamed
struct - should not be possible
is this related to the nacl change ?

[rsc, 2014-05-20 05:55:32]

I added the missing mkzip.go and testzip.proto and removed the spurious
cgo/gcc.go.

Thanks.

[dave_cheney.net, 2014-05-20 05:58:09]

Did you forget to hg add ../misc/mkzip.go ?

[dave_cheney.net, 2014-05-20 06:03:53]

Thanks, I'm now seeing this error on linux/amd64 with any binary.

SysMap(0xf01000, 0x1000): 0xffffffea
runtime: address space conflict: map(0xf01000) = 0xffffffea
fatal error: runtime: address space conflict

I'm using pepper_33

lucky(~/go/src) % file $(which sel_ldr_x86_32) $(which sel_ldr_x86_64)
/home/dfc/bin/sel_ldr_x86_32: symbolic link to
`nacl_sdk.d/pepper_33/tools/sel_ldr_x86_32'
/home/dfc/bin/sel_ldr_x86_64: symbolic link to
`nacl_sdk.d/pepper_33/tools/sel_ldr_x86_64'

On Tue, May 20, 2014 at 3:55 PM, Russ Cox <rsc@golang.org> wrote:
> I added the missing mkzip.go and testzip.proto and removed the spurious
> cgo/gcc.go.
>
> Thanks.
>

[rsc, 2014-05-20 06:05:49]

don't use pepper 33 then. :-)
i am using something like 27.
one thing at a time.

[dave_cheney.net, 2014-05-20 06:16:06]

> don't use pepper 33 then. :-)

Yup, that is fair. I downgraded to pepper_27 and still see the same
error. This is on linux/amd64 ubuntu 14.04, once my Mac has finished
digesting this load of builds, I'll try this CL there.

LGTM. I don't think this is reason enough not to land this change.
Tests pass on linux/amd64 and darwin/amd64 so this extra code is not
going to make nacl any worse.

You need to apply this small change because the three lines to skip
nacl in objdump_test.go have moved things down. Strangely this test
does not fail on darwin.

@@ -81,7 +89,7 @@
                t.Fatalf("objdump_test.go and %s are not same file", srcPath)
        }
        if srcLineNo != "89" {
-               t.Fatalf("line number = %v; want 89", srcLineNo)
+               t.Fatalf("line number = %v; want 97", srcLineNo)
        }

SysMap(0xf01000, 0x1000): 0xffffffea
runtime: address space conflict: map(0xf01000) = 0xffffffea
fatal error: runtime: address space conflict

goroutine 16 [running]:
runtime.throw(0xe6ea0e)
        /home/dfc/go/src/pkg/runtime/panic.c:520 +0xa0 fp=0x3eee1b10
runtime.SysMap(0xf01000, 0x1000, 0xffffff00, 0xe77e60)
        /home/dfc/go/src/pkg/runtime/mem_nacl.c:99 +0x1a0 fp=0x3eee1b30
runtime.MHeap_MapSpans(0xe82d40)
        /home/dfc/go/src/pkg/runtime/mheap.c:88 +0xa0 fp=0x3eee1b48
runtime.MHeap_SysAlloc(0xe82d40, 0x100000)
        /home/dfc/go/src/pkg/runtime/malloc.goc:618 +0x160 fp=0x3eee1b78
MHeap_Grow(0xe82d40, 0x60)
        /home/dfc/go/src/pkg/runtime/mheap.c:319 +0x60 fp=0x3eee1ba0
MHeap_AllocLocked(0xe82d40, 0x5e, 0x0)
        /home/dfc/go/src/pkg/runtime/mheap.c:222 +0x3c0 fp=0x3eee1bc0
runtime.MHeap_Alloc(0xe82d40, 0x5e, 0x0, 0x1)
        /home/dfc/go/src/pkg/runtime/mheap.c:178 +0xc0 fp=0x3eee1bd4
largealloc(0x9, 0x3eee1c2c)
        /home/dfc/go/src/pkg/runtime/malloc.goc:223 +0xc0 fp=0x3eee1bf8
runtime.mallocgc(0xbc000, 0x18dc61, 0x9)
        /home/dfc/go/src/pkg/runtime/malloc.goc:168 +0x100 fp=0x3eee1c2c
growslice1(0x1883a0, 0x116c0000, 0x96000, 0x96000, 0x96001, 0x3eee1c94)
        /home/dfc/go/src/pkg/runtime/slice.goc:125 +0x100 fp=0x3eee1c54
runtime.growslice(0x1883a0, 0x116c0000, 0x96000, 0x96000, 0x1, 0x0,
0x0, 0x0, 0x0)
        /home/dfc/go/src/pkg/runtime/slice.goc:76 +0x160 fp=0x3eee1c7c
syscall.(*decompressor).huffmanBlock(0x114ab000, 0x114ab020, 0x114ab0bc)
        /home/dfc/go/src/pkg/syscall/unzip_nacl.go:262 +0x2c0 fp=0x3eee1cf8
syscall.(*decompressor).nextBlock(0x114ab000)
        /home/dfc/go/src/pkg/syscall/unzip_nacl.go:55 +0x140 fp=0x3eee1d18
syscall.inflate(0x30ce98, 0xcfb7f, 0xdbc477, 0x29, 0x1)
        /home/dfc/go/src/pkg/syscall/unzip_nacl.go:503 +0x80 fp=0x3eee1d2c
syscall.unzip(0x22f6e8, 0xbab8e6)
        /home/dfc/go/src/pkg/syscall/unzip_nacl.go:621 +0x860 fp=0x3eee1ebc
syscall.init·3()
        /home/dfc/go/src/pkg/syscall/fstest_nacl.go:4 +0x40 fp=0x3eee1ec8
syscall.init()
        /home/dfc/go/src/pkg/syscall/zsyscall_nacl_386.go:63 +0x200
fp=0x3eee1ee4
time.init()
        /home/dfc/go/src/pkg/time/zoneinfo_unix.go:84 +0xc0 fp=0x3eee1f04
os.init()
        /home/dfc/go/src/pkg/os/types_notwin.go:25 +0xc0 fp=0x3eee1f20
fmt.init()
        /home/dfc/go/src/pkg/fmt/scan.go:1168 +0xe0 fp=0x3eee1f54
runtime/pprof.init()
        /home/dfc/go/src/pkg/runtime/pprof/pprof.go:679 +0x100 fp=0x3eee1f58
testing.init()
        /home/dfc/go/src/pkg/testing/testing.go:657 +0xa0 fp=0x3eee1f78
cmd/nm.init()
        /home/dfc/go/src/cmd/nm/nm_test.go:99 +0x80 fp=0x3eee1f98
main.init()
        cmd/nm/_test/_testmain.go:48 +0x60 fp=0x3eee1f9c
runtime.main()
        /home/dfc/go/src/pkg/runtime/proc.c:240 +0x160 fp=0x3eee1fd0
runtime.goexit()
        /home/dfc/go/src/pkg/runtime/proc.c:1445 fp=0x3eee1fd4
created by _rt0_go
        /home/dfc/go/src/pkg/runtime/asm_386.s:101 +0x1e0
FAIL    cmd/nm  0.235s

% file $(which sel_ldr_x86_32) $(which sel_ldr_x86_64)
/home/dfc/bin/sel_ldr_x86_32: symbolic link to
`nacl_sdk.d/pepper_27/tools/sel_ldr_x86_32'
/home/dfc/bin/sel_ldr_x86_64: symbolic link to
`nacl_sdk.d/pepper_27/tools/sel_ldr_x86_64'

[dave_cheney.net, 2014-05-20 06:18:56]

Sorry, make that patch

@@ -80,8 +88,8 @@
        if !os.SameFile(fi1, fi2) {
                t.Fatalf("objdump_test.go and %s are not same file", srcPath)
        }
-       if srcLineNo != "89" {
-               t.Fatalf("line number = %v; want 89", srcLineNo)
+       if srcLineNo != "97" {
+               t.Fatalf("line number = %v; want 97", srcLineNo)
        }
 }

On Tue, May 20, 2014 at 4:16 PM, Dave Cheney <dave@cheney.net> wrote:
>> don't use pepper 33 then. :-)
>
> Yup, that is fair. I downgraded to pepper_27 and still see the same
> error. This is on linux/amd64 ubuntu 14.04, once my Mac has finished
> digesting this load of builds, I'll try this CL there.
>
> LGTM. I don't think this is reason enough not to land this change.
> Tests pass on linux/amd64 and darwin/amd64 so this extra code is not
> going to make nacl any worse.
>
> You need to apply this small change because the three lines to skip
> nacl in objdump_test.go have moved things down. Strangely this test
> does not fail on darwin.
>
> @@ -81,7 +89,7 @@
>                 t.Fatalf("objdump_test.go and %s are not same file", srcPath)
>         }
>         if srcLineNo != "89" {
> -               t.Fatalf("line number = %v; want 89", srcLineNo)
> +               t.Fatalf("line number = %v; want 97", srcLineNo)
>         }
>
> SysMap(0xf01000, 0x1000): 0xffffffea
> runtime: address space conflict: map(0xf01000) = 0xffffffea
> fatal error: runtime: address space conflict
>
> goroutine 16 [running]:
> runtime.throw(0xe6ea0e)
>         /home/dfc/go/src/pkg/runtime/panic.c:520 +0xa0 fp=0x3eee1b10
> runtime.SysMap(0xf01000, 0x1000, 0xffffff00, 0xe77e60)
>         /home/dfc/go/src/pkg/runtime/mem_nacl.c:99 +0x1a0 fp=0x3eee1b30
> runtime.MHeap_MapSpans(0xe82d40)
>         /home/dfc/go/src/pkg/runtime/mheap.c:88 +0xa0 fp=0x3eee1b48
> runtime.MHeap_SysAlloc(0xe82d40, 0x100000)
>         /home/dfc/go/src/pkg/runtime/malloc.goc:618 +0x160 fp=0x3eee1b78
> MHeap_Grow(0xe82d40, 0x60)
>         /home/dfc/go/src/pkg/runtime/mheap.c:319 +0x60 fp=0x3eee1ba0
> MHeap_AllocLocked(0xe82d40, 0x5e, 0x0)
>         /home/dfc/go/src/pkg/runtime/mheap.c:222 +0x3c0 fp=0x3eee1bc0
> runtime.MHeap_Alloc(0xe82d40, 0x5e, 0x0, 0x1)
>         /home/dfc/go/src/pkg/runtime/mheap.c:178 +0xc0 fp=0x3eee1bd4
> largealloc(0x9, 0x3eee1c2c)
>         /home/dfc/go/src/pkg/runtime/malloc.goc:223 +0xc0 fp=0x3eee1bf8
> runtime.mallocgc(0xbc000, 0x18dc61, 0x9)
>         /home/dfc/go/src/pkg/runtime/malloc.goc:168 +0x100 fp=0x3eee1c2c
> growslice1(0x1883a0, 0x116c0000, 0x96000, 0x96000, 0x96001, 0x3eee1c94)
>         /home/dfc/go/src/pkg/runtime/slice.goc:125 +0x100 fp=0x3eee1c54
> runtime.growslice(0x1883a0, 0x116c0000, 0x96000, 0x96000, 0x1, 0x0,
> 0x0, 0x0, 0x0)
>         /home/dfc/go/src/pkg/runtime/slice.goc:76 +0x160 fp=0x3eee1c7c
> syscall.(*decompressor).huffmanBlock(0x114ab000, 0x114ab020, 0x114ab0bc)
>         /home/dfc/go/src/pkg/syscall/unzip_nacl.go:262 +0x2c0 fp=0x3eee1cf8
> syscall.(*decompressor).nextBlock(0x114ab000)
>         /home/dfc/go/src/pkg/syscall/unzip_nacl.go:55 +0x140 fp=0x3eee1d18
> syscall.inflate(0x30ce98, 0xcfb7f, 0xdbc477, 0x29, 0x1)
>         /home/dfc/go/src/pkg/syscall/unzip_nacl.go:503 +0x80 fp=0x3eee1d2c
> syscall.unzip(0x22f6e8, 0xbab8e6)
>         /home/dfc/go/src/pkg/syscall/unzip_nacl.go:621 +0x860 fp=0x3eee1ebc
> syscall.init·3()
>         /home/dfc/go/src/pkg/syscall/fstest_nacl.go:4 +0x40 fp=0x3eee1ec8
> syscall.init()
>         /home/dfc/go/src/pkg/syscall/zsyscall_nacl_386.go:63 +0x200
> fp=0x3eee1ee4
> time.init()
>         /home/dfc/go/src/pkg/time/zoneinfo_unix.go:84 +0xc0 fp=0x3eee1f04
> os.init()
>         /home/dfc/go/src/pkg/os/types_notwin.go:25 +0xc0 fp=0x3eee1f20
> fmt.init()
>         /home/dfc/go/src/pkg/fmt/scan.go:1168 +0xe0 fp=0x3eee1f54
> runtime/pprof.init()
>         /home/dfc/go/src/pkg/runtime/pprof/pprof.go:679 +0x100 fp=0x3eee1f58
> testing.init()
>         /home/dfc/go/src/pkg/testing/testing.go:657 +0xa0 fp=0x3eee1f78
> cmd/nm.init()
>         /home/dfc/go/src/cmd/nm/nm_test.go:99 +0x80 fp=0x3eee1f98
> main.init()
>         cmd/nm/_test/_testmain.go:48 +0x60 fp=0x3eee1f9c
> runtime.main()
>         /home/dfc/go/src/pkg/runtime/proc.c:240 +0x160 fp=0x3eee1fd0
> runtime.goexit()
>         /home/dfc/go/src/pkg/runtime/proc.c:1445 fp=0x3eee1fd4
> created by _rt0_go
>         /home/dfc/go/src/pkg/runtime/asm_386.s:101 +0x1e0
> FAIL    cmd/nm  0.235s
>
> % file $(which sel_ldr_x86_32) $(which sel_ldr_x86_64)
> /home/dfc/bin/sel_ldr_x86_32: symbolic link to
> `nacl_sdk.d/pepper_27/tools/sel_ldr_x86_32'
> /home/dfc/bin/sel_ldr_x86_64: symbolic link to
> `nacl_sdk.d/pepper_27/tools/sel_ldr_x86_64'

[dave_cheney.net, 2014-05-20 06:42:27]

LGTM. much closer using darwin/amd64 as my host

Only a few failures

** Signal 4 from untrusted code: pc=2fb1b
FAIL html/template 1.480s
** Signal 10 from untrusted code: pc=3dbee
FAIL net/http 6.033s
** Signal 4 from untrusted code: pc=2f9fb
FAIL sync/atomic 1.988s
** Signal 4 from untrusted code: pc=2fc7b
FAIL time 3.354s

On Tue, May 20, 2014 at 4:18 PM, Dave Cheney <dave@cheney.net> wrote:
> Sorry, make that patch
>
> @@ -80,8 +88,8 @@
>         if !os.SameFile(fi1, fi2) {
>                 t.Fatalf("objdump_test.go and %s are not same file", srcPath)
>         }
> -       if srcLineNo != "89" {
> -               t.Fatalf("line number = %v; want 89", srcLineNo)
> +       if srcLineNo != "97" {
> +               t.Fatalf("line number = %v; want 97", srcLineNo)
>         }
>  }
>
> On Tue, May 20, 2014 at 4:16 PM, Dave Cheney <dave@cheney.net> wrote:
>>> don't use pepper 33 then. :-)
>>
>> Yup, that is fair. I downgraded to pepper_27 and still see the same
>> error. This is on linux/amd64 ubuntu 14.04, once my Mac has finished
>> digesting this load of builds, I'll try this CL there.
>>
>> LGTM. I don't think this is reason enough not to land this change.
>> Tests pass on linux/amd64 and darwin/amd64 so this extra code is not
>> going to make nacl any worse.
>>
>> You need to apply this small change because the three lines to skip
>> nacl in objdump_test.go have moved things down. Strangely this test
>> does not fail on darwin.
>>
>> @@ -81,7 +89,7 @@
>>                 t.Fatalf("objdump_test.go and %s are not same file",
srcPath)
>>         }
>>         if srcLineNo != "89" {
>> -               t.Fatalf("line number = %v; want 89", srcLineNo)
>> +               t.Fatalf("line number = %v; want 97", srcLineNo)
>>         }
>>
>> SysMap(0xf01000, 0x1000): 0xffffffea
>> runtime: address space conflict: map(0xf01000) = 0xffffffea
>> fatal error: runtime: address space conflict
>>
>> goroutine 16 [running]:
>> runtime.throw(0xe6ea0e)
>>         /home/dfc/go/src/pkg/runtime/panic.c:520 +0xa0 fp=0x3eee1b10
>> runtime.SysMap(0xf01000, 0x1000, 0xffffff00, 0xe77e60)
>>         /home/dfc/go/src/pkg/runtime/mem_nacl.c:99 +0x1a0 fp=0x3eee1b30
>> runtime.MHeap_MapSpans(0xe82d40)
>>         /home/dfc/go/src/pkg/runtime/mheap.c:88 +0xa0 fp=0x3eee1b48
>> runtime.MHeap_SysAlloc(0xe82d40, 0x100000)
>>         /home/dfc/go/src/pkg/runtime/malloc.goc:618 +0x160 fp=0x3eee1b78
>> MHeap_Grow(0xe82d40, 0x60)
>>         /home/dfc/go/src/pkg/runtime/mheap.c:319 +0x60 fp=0x3eee1ba0
>> MHeap_AllocLocked(0xe82d40, 0x5e, 0x0)
>>         /home/dfc/go/src/pkg/runtime/mheap.c:222 +0x3c0 fp=0x3eee1bc0
>> runtime.MHeap_Alloc(0xe82d40, 0x5e, 0x0, 0x1)
>>         /home/dfc/go/src/pkg/runtime/mheap.c:178 +0xc0 fp=0x3eee1bd4
>> largealloc(0x9, 0x3eee1c2c)
>>         /home/dfc/go/src/pkg/runtime/malloc.goc:223 +0xc0 fp=0x3eee1bf8
>> runtime.mallocgc(0xbc000, 0x18dc61, 0x9)
>>         /home/dfc/go/src/pkg/runtime/malloc.goc:168 +0x100 fp=0x3eee1c2c
>> growslice1(0x1883a0, 0x116c0000, 0x96000, 0x96000, 0x96001, 0x3eee1c94)
>>         /home/dfc/go/src/pkg/runtime/slice.goc:125 +0x100 fp=0x3eee1c54
>> runtime.growslice(0x1883a0, 0x116c0000, 0x96000, 0x96000, 0x1, 0x0,
>> 0x0, 0x0, 0x0)
>>         /home/dfc/go/src/pkg/runtime/slice.goc:76 +0x160 fp=0x3eee1c7c
>> syscall.(*decompressor).huffmanBlock(0x114ab000, 0x114ab020, 0x114ab0bc)
>>         /home/dfc/go/src/pkg/syscall/unzip_nacl.go:262 +0x2c0 fp=0x3eee1cf8
>> syscall.(*decompressor).nextBlock(0x114ab000)
>>         /home/dfc/go/src/pkg/syscall/unzip_nacl.go:55 +0x140 fp=0x3eee1d18
>> syscall.inflate(0x30ce98, 0xcfb7f, 0xdbc477, 0x29, 0x1)
>>         /home/dfc/go/src/pkg/syscall/unzip_nacl.go:503 +0x80 fp=0x3eee1d2c
>> syscall.unzip(0x22f6e8, 0xbab8e6)
>>         /home/dfc/go/src/pkg/syscall/unzip_nacl.go:621 +0x860 fp=0x3eee1ebc
>> syscall.init·3()
>>         /home/dfc/go/src/pkg/syscall/fstest_nacl.go:4 +0x40 fp=0x3eee1ec8
>> syscall.init()
>>         /home/dfc/go/src/pkg/syscall/zsyscall_nacl_386.go:63 +0x200
>> fp=0x3eee1ee4
>> time.init()
>>         /home/dfc/go/src/pkg/time/zoneinfo_unix.go:84 +0xc0 fp=0x3eee1f04
>> os.init()
>>         /home/dfc/go/src/pkg/os/types_notwin.go:25 +0xc0 fp=0x3eee1f20
>> fmt.init()
>>         /home/dfc/go/src/pkg/fmt/scan.go:1168 +0xe0 fp=0x3eee1f54
>> runtime/pprof.init()
>>         /home/dfc/go/src/pkg/runtime/pprof/pprof.go:679 +0x100 fp=0x3eee1f58
>> testing.init()
>>         /home/dfc/go/src/pkg/testing/testing.go:657 +0xa0 fp=0x3eee1f78
>> cmd/nm.init()
>>         /home/dfc/go/src/cmd/nm/nm_test.go:99 +0x80 fp=0x3eee1f98
>> main.init()
>>         cmd/nm/_test/_testmain.go:48 +0x60 fp=0x3eee1f9c
>> runtime.main()
>>         /home/dfc/go/src/pkg/runtime/proc.c:240 +0x160 fp=0x3eee1fd0
>> runtime.goexit()
>>         /home/dfc/go/src/pkg/runtime/proc.c:1445 fp=0x3eee1fd4
>> created by _rt0_go
>>         /home/dfc/go/src/pkg/runtime/asm_386.s:101 +0x1e0
>> FAIL    cmd/nm  0.235s
>>
>> % file $(which sel_ldr_x86_32) $(which sel_ldr_x86_64)
>> /home/dfc/bin/sel_ldr_x86_32: symbolic link to
>> `nacl_sdk.d/pepper_27/tools/sel_ldr_x86_32'
>> /home/dfc/bin/sel_ldr_x86_64: symbolic link to
>> `nacl_sdk.d/pepper_27/tools/sel_ldr_x86_64'

[bradfitz, 2014-05-20 06:47:25]

https://codereview.appspot.com/100590044/diff/90001/misc/nacl/mkzip.go
File misc/nacl/mkzip.go (right):

https://codereview.appspot.com/100590044/diff/90001/misc/nacl/mkzip.go#newcode76
misc/nacl/mkzip.go:76: defer gw.Close()
ignoring close of a file opened for write

https://codereview.appspot.com/100590044/diff/90001/misc/nacl/mkzip.go#newcod...
misc/nacl/mkzip.go:207: fmt.Fprintf(w.b, "\\x%02x", c)
ignoring error

https://codereview.appspot.com/100590044/diff/90001/misc/nacl/mkzip.go#newcod...
misc/nacl/mkzip.go:213: fmt.Fprintf(w.b, "\")\n}\n")
ignoring error, and the next line

https://codereview.appspot.com/100590044/diff/90001/src/pkg/net/timeout_test.go
File src/pkg/net/timeout_test.go (right):

https://codereview.appspot.com/100590044/diff/90001/src/pkg/net/timeout_test....
src/pkg/net/timeout_test.go:123: if err == io.EOF && runtime.GOOS == "nacl" { //
close enough
at least file a bug and reference it here? otherwise why are we testing it here?
this isn't one of those things like "nacl can't do child processes".

https://codereview.appspot.com/100590044/diff/90001/src/run.bash
File src/run.bash (right):

https://codereview.appspot.com/100590044/diff/90001/src/run.bash#newcode181
src/run.bash:181: [ "$GOOS" = nacl ] ||
as below...

https://codereview.appspot.com/100590044/diff/90001/src/run.bash#newcode186
src/run.bash:186: [ "$GOOS" = nacl ] ||
Why is this line = but the next line ==?

https://codereview.appspot.com/100590044/diff/90001/src/run.bash#newcode192
src/run.bash:192: [ "$GOOS" = nacl ] ||
etc

https://codereview.appspot.com/100590044/diff/90001/src/run.bash#newcode202
src/run.bash:202: [ "$GOOS" = nacl ] ||
etc

https://codereview.appspot.com/100590044/diff/90001/src/run.bash#newcode222
src/run.bash:222: [ "$GOOS" == nacl ] ||
but this one is == ?

[bradfitz, 2014-05-20 15:09:08]

https://codereview.appspot.com/100590044/diff/90001/src/nacl.bash
File src/nacl.bash (right):

https://codereview.appspot.com/100590044/diff/90001/src/nacl.bash#newcode10
src/nacl.bash:10: # Assumes that sel_ldr binaries are in $PATH; see
../misc/nacl/README.
can you check that they are found in $PATH before running make.bash, to fail
faster earlier?

[rsc, 2014-05-20 15:53:25]

https://codereview.appspot.com/100590044/diff/90001/misc/nacl/mkzip.go
File misc/nacl/mkzip.go (right):

https://codereview.appspot.com/100590044/diff/90001/misc/nacl/mkzip.go#newcode76
misc/nacl/mkzip.go:76: defer gw.Close()
On 2014/05/20 06:47:24, bradfitz wrote:
> ignoring close of a file opened for write

Done.

https://codereview.appspot.com/100590044/diff/90001/src/nacl.bash
File src/nacl.bash (right):

https://codereview.appspot.com/100590044/diff/90001/src/nacl.bash#newcode1
src/nacl.bash:1: #!/bin/bash
FWIW I renamed this to nacltest.bash to make clear that you do not use it to
build the standard nacl toolchain.

https://codereview.appspot.com/100590044/diff/90001/src/nacl.bash#newcode10
src/nacl.bash:10: # Assumes that sel_ldr binaries are in $PATH; see
../misc/nacl/README.
On 2014/05/20 15:09:08, bradfitz wrote:
> can you check that they are found in $PATH before running make.bash, to fail
> faster earlier?

Done.

https://codereview.appspot.com/100590044/diff/90001/src/pkg/net/timeout_test.go
File src/pkg/net/timeout_test.go (right):

https://codereview.appspot.com/100590044/diff/90001/src/pkg/net/timeout_test....
src/pkg/net/timeout_test.go:123: if err == io.EOF && runtime.GOOS == "nacl" { //
close enough
On 2014/05/20 06:47:24, bradfitz wrote:
> at least file a bug and reference it here? otherwise why are we testing it
here?
> this isn't one of those things like "nacl can't do child processes".

Done.

https://codereview.appspot.com/100590044/diff/90001/src/run.bash
File src/run.bash (right):

https://codereview.appspot.com/100590044/diff/90001/src/run.bash#newcode181
src/run.bash:181: [ "$GOOS" = nacl ] ||
On 2014/05/20 06:47:25, bradfitz wrote:
> as below...

Done.

https://codereview.appspot.com/100590044/diff/90001/src/run.bash#newcode186
src/run.bash:186: [ "$GOOS" = nacl ] ||
On 2014/05/20 06:47:24, bradfitz wrote:
> Why is this line = but the next line ==?

Done. (I have no idea.)

https://codereview.appspot.com/100590044/diff/90001/src/run.bash#newcode192
src/run.bash:192: [ "$GOOS" = nacl ] ||
On 2014/05/20 06:47:24, bradfitz wrote:
> etc

Done.

https://codereview.appspot.com/100590044/diff/90001/src/run.bash#newcode202
src/run.bash:202: [ "$GOOS" = nacl ] ||
On 2014/05/20 06:47:24, bradfitz wrote:
> etc

Done.

https://codereview.appspot.com/100590044/diff/90001/src/run.bash#newcode222
src/run.bash:222: [ "$GOOS" == nacl ] ||
On 2014/05/20 06:47:24, bradfitz wrote:
> but this one is == ?

Done.

[rsc, 2014-05-20 16:10:21]

*** Submitted as https://code.google.com/p/go/source/detail?r=29348a1c6914 ***

build: make nacl pass

Add nacl.bash, the NaCl version of all.bash.
It's a separate script because it builds a variant of package syscall
with a large zip file embedded in it, containing all the input files
needed for tests.

Disable various tests new since the last round, mostly the ones using os/exec.

Fixes issue 7945.

LGTM=dave
R=golang-codereviews, remyoudompheng, dave, bradfitz
CC=golang-codereviews
https://codereview.appspot.com/100590044