os/exec: TestPipeLookPathLeak fails randomly on OS X

[davecheney]

Darwin builders have started to occasionally fail this test

--- FAIL: TestPipeLookPathLeak (0.02 seconds)
    exec_test.go:169: leaked 4 fds; want ~0
FAIL
FAIL    os/exec 1.083s


Recent failures:
http://build.golang.org/log/5b9a4be800936ee52a6c00b1b64337ae9d4e1906
http://build.golang.org/log/5fa2f5674bc6293bf068689b4df3e3baf96810e1

[davecheney]

Comment 1:

I can confirm this test failure, or more correctly, the underlying leak is present. If I
modify the finaliser attached to each os.File to complain if it is being called, ie, if
the os.File has fallen out of scope without a call to Close() and stress test this
package eventually I find a stray os.File (attached to a tcp connection) which is
probably what this test is also discovering.
At this point I haven't figured out where we are leaking the os.File (or net.Conn)

Labels changed: added priority-soon, removed priority-triage.

Owner changed to @davecheney.

[davecheney]

Comment 2:

http://build.golang.org/log/f88e8e657f72668117815e9b5a5973d1e3bf56d1

[rsc]

Comment 3:

Labels changed: added go1.2.

[davecheney]

Comment 4:

Removing myself as the owner. Reviewing the failures, I think they are all for 10.6,
which points to a kernel bug.

Owner changed to ---.

[rsc]

Comment 5:

Labels changed: added testing.

[rsc]

Comment 6:

Labels changed: added go1.2maybe, removed go1.2.

[rsc]

Comment 7:

Labels changed: added go1.3, removed go1.2maybe.

[rsc]

Comment 8:

Labels changed: added release-go1.3.

[rsc]

Comment 9:

Labels changed: removed go1.3.

[rsc]

Comment 10:

Labels changed: added repo-main.

[rsc]

Comment 11:

Labels changed: added priority-later, release-none, suggested, removed priority-soon, release-go1.3.

[rsc]

Per #9511, we will not be making any further bug fixes specific to 10.6.