code review 7545043: fmt: Skip TestCountMallocs if GOMAXPROCS>1.

all: Skip AllocsPerRun tests if GOMAXPROCS>1.

Fixes issue 4974.

[albert.strasheim, 2013-03-06 23:10:20]

Hello rsc@golang.org (cc: golang-dev@googlegroups.com),

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

[bradfitz, 2013-03-06 23:13:15]

LGTM

On Wed, Mar 6, 2013 at 3:10 PM, <fullung@gmail.com> wrote:

> Reviewers: rsc,
>
> Message:
> Hello rsc@golang.org (cc: golang-dev@googlegroups.com),
>
> I'd like you to review this change to
> https://code.google.com/p/go/
>
>
> Description:
> fmt: Skip TestCountMallocs if GOMAXPROCS>1.
>
> Fixes issue 4974.
>
> Please review this at
https://codereview.appspot.**com/7545043/<https://codereview.appspot.com/7545...
>
> Affected files:
>   M src/pkg/fmt/fmt_test.go
>
>
> Index: src/pkg/fmt/fmt_test.go
> ==============================**==============================**=======
> --- a/src/pkg/fmt/fmt_test.go
> +++ b/src/pkg/fmt/fmt_test.go
> @@ -9,6 +9,7 @@
>         . "fmt"
>         "io"
>         "math"
> +       "runtime"
>         "strings"
>         "testing"
>         "time"
> @@ -601,6 +602,9 @@
>  var _ bytes.Buffer
>
>  func TestCountMallocs(t *testing.T) {
> +       if runtime.GOMAXPROCS(0) > 1 {
> +               t.Skip("skipping; GOMAXPROCS>1")
> +       }
>         for _, mt := range mallocTest {
>                 mallocs := testing.AllocsPerRun(100, mt.fn)
>                 if got, max := mallocs, float64(mt.count); got > max {
>
>
> --
>
> ---You received this message because you are subscribed to the Google
> Groups "golang-dev" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to
golang-dev+unsubscribe@**googlegroups.com<golang-dev%2Bunsubscribe@googlegrou...
> .
> For more options, visit
https://groups.google.com/**groups/opt_out<https://groups.google.com/groups/o...
> .
>
>
>

[r, 2013-03-06 23:13:29]

What about these?

src/pkg/encoding/gob/timing_test.go:50:func TestCountEncodeMallocs(t
*testing.T) {
src/pkg/encoding/gob/timing_test.go:66:func TestCountDecodeMallocs(t
*testing.T) {
src/pkg/net/http/header_test.go:193:func
TestHeaderWriteSubsetMallocs(t *testing.T) {
src/pkg/net/rpc/server_test.go:466:func TestCountMallocs(t *testing.T) {
src/pkg/net/rpc/server_test.go:470:func TestCountMallocsOverHTTP(t *testing.T)
{
src/pkg/strconv/strconv_test.go:44:func TestCountMallocs(t *testing.T) {
src/pkg/time/time_test.go:1300:func TestCountMallocs(t *testing.T) {

[albert.strasheim, 2013-03-06 23:14:37]

On 2013/03/06 23:13:29, r wrote:
> What about these?

I'll take a quick look and report back.

[albert.strasheim, 2013-03-06 23:34:15]

PTAL

I've never seen any of the other ones fail, but I guess it's only a matter of
time.

[r, 2013-03-06 23:50:09]

LGTM

[r, 2013-03-06 23:52:34]

*** Submitted as https://code.google.com/p/go/source/detail?r=cb98d0bcf923 ***

all: Skip AllocsPerRun tests if GOMAXPROCS>1.

Fixes issue 4974.

R=rsc, bradfitz, r
CC=golang-dev
https://codereview.appspot.com/7545043

Committer: Rob Pike <r@golang.org>

[minux1, 2013-03-07 10:03:32]

this is strange, testing.AllocsPerRun already force GOMAXPROCS to 1
before tests start.

perhaps this is a scheduler bug than a test bug.

[rsc, 2013-03-07 14:18:37]

On Thu, Mar 7, 2013 at 5:03 AM, minux <minux.ma@gmail.com> wrote:

> this is strange, testing.AllocsPerRun already force GOMAXPROCS to 1
> before tests start.
>
> perhaps this is a scheduler bug than a test bug.
>

Want to take a look? It was failing with 1.2 allocations per test instead
of the expected 1.0.

Russ

[minux1, 2013-03-07 14:44:12]

On Thu, Mar 7, 2013 at 10:18 PM, Russ Cox <rsc@golang.org> wrote:
> Want to take a look? It was failing with 1.2 allocations per test instead of
> the expected 1.0.
I'm still tackling the cgo problems on arm,
so i filed https://code.google.com/p/go/issues/detail?id=5000 to remind us.

Note:
from the original issue report, fullung said that the allocation comes from
runtime, so we should do something or this problem might affect other
Go 1 allocation tests in the wild that adopt the idiom:
defer runtime.GOMAXPROCS(runtime.GOMAXPROCS(1))
before tests.