code review 102710043: runtime: fix nacl amd64p32 flakiness

runtime: fix nacl amd64p32 flakiness

newproc takes two extra pointers, not two extra registers.
On amd64p32 (nacl) they are different.

We diagnosed this before the 1.3 cut but the tree was frozen.
I believe this is causing the random problems on the builder.

Fixes issue 8199.

[rsc, 2014-06-28 00:13:13]

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

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

[rsc, 2014-06-28 00:13:19]

*** Submitted as https://code.google.com/p/go/source/detail?r=5640e19d768d ***

runtime: fix nacl amd64p32 flakiness

newproc takes two extra pointers, not two extra registers.
On amd64p32 (nacl) they are different.

We diagnosed this before the 1.3 cut but the tree was frozen.
I believe this is causing the random problems on the builder.

Fixes issue 8199.

TBR=r
CC=golang-codereviews
https://codereview.appspot.com/102710043

[dave_cheney.net, 2014-06-28 00:19:19]

LGTM. 

On 28/06/2014, at 10:13, rsc@golang.org wrote:

> Reviewers: r,
> 
> Message:
> Hello r (cc: golang-codereviews@googlegroups.com),
> 
> I'd like you to review this change to
> https://code.google.com/p/go/
> 
> 
> Description:
> runtime: fix nacl amd64p32 flakiness
> 
> newproc takes two extra pointers, not two extra registers.
> On amd64p32 (nacl) they are different.
> 
> We diagnosed this before the 1.3 cut but the tree was frozen.
> I believe this is causing the random problems on the builder.
> 
> Fixes issue 8199.
> 
> Please review this at https://codereview.appspot.com/102710043/
> 
> Affected files (+1, -1 lines):
>  M src/pkg/runtime/traceback_x86.c
> 
> 
> Index: src/pkg/runtime/traceback_x86.c
> ===================================================================
> --- a/src/pkg/runtime/traceback_x86.c
> +++ b/src/pkg/runtime/traceback_x86.c
> @@ -214,7 +214,7 @@
>        // the SP is two words lower than normal.
>        sparg = frame.sp;
>        if(wasnewproc)
> -            sparg += 2*sizeof(uintreg);
> +            sparg += 2*sizeof(uintptr);
> 
>        // Determine frame's 'continuation PC', where it can continue.
>        // Normally this is the return address on the stack, but if sigpanic
> 
> 
> -- 
> You received this message because you are subscribed to the Google Groups
"golang-codereviews" group.
> To unsubscribe from this group and stop receiving emails from it, send an
email to golang-codereviews+unsubscribe@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

[gobot, 2014-06-28 00:32:21]

This CL appears to have broken the linux-arm-cheney-panda builder.
See http://build.golang.org/log/72c3bd3b629a5c6eb0ca36338c0c26854d26997c

[dave_cheney.net, 2014-06-28 01:20:50]

It could be that this builder, after 3+ years of faithful service is
on its last legs, but I'd be willing to place a small side bet that
the recent removal of register m could be related

# GOMAXPROCS=2 runtime -cpu=1,2,4
--- FAIL: TestCgoCrashHandler-2 (1.33s)
crash_test.go:85: output:
runtime: newstack called from g=0x10732640�����<�������� ��
panic
during panic

runtime stack:
runtime.startpanic()
/usr/local/go/src/pkg/runtime/panic.c:408 +0xc4
runtime.throw(0x523963)
/usr/local/go/src/pkg/runtime/panic.c:520 +0x3c
acquirep(0x10714500)
/usr/local/go/src/pkg/runtime/proc.c:2450 +0x50
exitsyscall0(0x10703040)
/usr/local/go/src/pkg/runtime/proc.c:1695 +0x94
runtime.mcall(0x10714500)
/usr/local/go/src/pkg/runtime/asm_arm.s:168 +0x50

goroutine 16 [semacquire]:
sync.runtime_Semacquire(0x10886650)
/usr/local/go/src/pkg/runtime/sema.goc:199 +0x34
sync.(*WaitGroup).Wait(0x10852860)
/usr/local/go/src/pkg/sync/waitgroup.go:129 +0x174
main.(*builder).do(0x107e1880, 0x107f4af0)
/usr/local/go/src/cmd/go/build.go:760 +0x458
main.runRun(0x51e2a8, 0x1070e010, 0x1, 0x1)
/usr/local/go/src/cmd/go/run.go:115 +0x7dc
main.main()
/usr/local/go/src/cmd/go/main.go:162 +0x518

On Sat, Jun 28, 2014 at 10:32 AM,  <gobot@golang.org> wrote:
> This CL appears to have broken the linux-arm-cheney-panda builder.
> See http://build.golang.org/log/72c3bd3b629a5c6eb0ca36338c0c26854d26997c
>
> https://codereview.appspot.com/102710043/

[josharian, 2014-06-28 22:30:34]

Filed issue 8298.


On Fri, Jun 27, 2014 at 6:20 PM, Dave Cheney <dave@cheney.net> wrote:

> It could be that this builder, after 3+ years of faithful service is
> on its last legs, but I'd be willing to place a small side bet that
> the recent removal of register m could be related
>
> # GOMAXPROCS=2 runtime -cpu=1,2,4
> --- FAIL: TestCgoCrashHandler-2 (1.33s)
> crash_test.go:85: output:
> runtime: newstack called from g=0x10732640�� ���< ������  ��  �� panic
> during panic
>
> runtime stack:
> runtime.startpanic()
> /usr/local/go/src/pkg/runtime/panic.c:408 +0xc4
> runtime.throw(0x523963)
> /usr/local/go/src/pkg/runtime/panic.c:520 +0x3c
> acquirep(0x10714500)
> /usr/local/go/src/pkg/runtime/proc.c:2450 +0x50
> exitsyscall0(0x10703040)
> /usr/local/go/src/pkg/runtime/proc.c:1695 +0x94
> runtime.mcall(0x10714500)
> /usr/local/go/src/pkg/runtime/asm_arm.s:168 +0x50
>
> goroutine 16 [semacquire]:
> sync.runtime_Semacquire(0x10886650)
> /usr/local/go/src/pkg/runtime/sema.goc:199 +0x34
> sync.(*WaitGroup).Wait(0x10852860)
> /usr/local/go/src/pkg/sync/waitgroup.go:129 +0x174
> main.(*builder).do(0x107e1880, 0x107f4af0)
> /usr/local/go/src/cmd/go/build.go:760 +0x458
> main.runRun(0x51e2a8, 0x1070e010, 0x1, 0x1)
> /usr/local/go/src/cmd/go/run.go:115 +0x7dc
> main.main()
> /usr/local/go/src/cmd/go/main.go:162 +0x518
>
> On Sat, Jun 28, 2014 at 10:32 AM,  <gobot@golang.org> wrote:
> > This CL appears to have broken the linux-arm-cheney-panda builder.
> > See http://build.golang.org/log/72c3bd3b629a5c6eb0ca36338c0c26854d26997c
> >
> > https://codereview.appspot.com/102710043/
>
> --
> You received this message because you are subscribed to the Google Groups
> "golang-codereviews" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to golang-codereviews+unsubscribe@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
>