On 03/11/2016 13:53, Radim Krčmář wrote:
> 2016-11-03 13:38+0100, Paolo Bonzini:
>> On 02/11/2016 22:05, Andrew Jones wrote:
>>> On Wed, Nov 02, 2016 at 09:50:29PM +0100, Radim Krčmář wrote:
>>>> Not doing so hid stderr output when the test was stuck and subsequently
>>>> killed with ^C.
>>>>
>>>> Signed-off-by: Radim Krčmář <rkrcmar@xxxxxxxxxx>
>>>> ---
>>>> scripts/arch-run.bash | 3 +--
>>>> 1 file changed, 1 insertion(+), 2 deletions(-)
>>>>
>>>> diff --git a/scripts/arch-run.bash b/scripts/arch-run.bash
>>>> index 8e75c6ed6e17..326f59484e2b 100644
>>>> --- a/scripts/arch-run.bash
>>>> +++ b/scripts/arch-run.bash
>>>> @@ -30,12 +30,11 @@ run_qemu ()
>>>>
>>>> # stdout to {stdout}, stderr to $errors
>>>> exec {stdout}>&1
>>>> - errors=$("${@}" 2>&1 1>&${stdout})
>>>> + errors=$("${@}" 2> >(tee >(cat) >&2) 1>&${stdout})
>>
>> What about:
>>
>> # preserve stdout and stderr output, but save stderr to $errors
>> exec {stdout}>&1 {stderr}>&2
>> errors=$("$@" 2> >(tee /dev/fd/$stderr) > /dev/fd/$stdout)
>> exec {stdout}>&- {stderr}>&-
>>
>> (Tested in the shell but not in arch-run.bash).
>
> Using files looks nicer, but is there a problem in the redirection?
> I would rather use /dev/stderr or /dev/fd/2 directly if not.
>
> i.e.
> exec {stdout}>&1
> errors=$("$@" 2> >(tee /dev/stderr) > /dev/fd/$stdout)
> exec {stdout}>&-
Yes, this works too.
Paolo
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
