GNU bug report logs - #14897
dd skip exit code

Previous Next

To add a comment to this bug, you must first unarchive it, by sending
a message to control AT debbugs.gnu.org, with unarchive 14897 in the body.
You can then email your comments to 14897 AT debbugs.gnu.org in the normal way.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: ffx <ffx-gdev <at> tfq.fr>
To: bug-coreutils <at> gnu.org
Subject: dd skip exit code
Date: Thu, 18 Jul 2013 11:03:38 +0200

Hello

here is a small test I did

  bash$ > foo

  bash$ dd if=foo skip=10
  dd: `foo': cannot skip to specified offset
  0+0 records in
  0+0 records out
  0 bytes (0 B) copied, 0.000167556 s, 0.0 kB/s

  bash$ echo $?
  0

  bash$ dd if=bar skip=10
  dd: opening `bar': No such file or directory

  bash$ echo $?
  1

The test with bar is only to check that exit code can be set.
Is it the expected behaviour to have a success exit code when skip fails, even with the notice message "cannot skip..."?
I would rather expect dd to exit with an error code for the failed skip.

-- 
ffx

Message #8 received at 14897 <at> debbugs.gnu.org (full text, mbox):

From: Paul Eggert <eggert <at> cs.ucla.edu>
To: ffx <ffx-gdev <at> tfq.fr>
Cc: 14897 <at> debbugs.gnu.org
Subject: Re: bug#14897: dd skip exit code
Date: Thu, 18 Jul 2013 07:40:31 -0700

On 07/18/2013 02:03 AM, ffx wrote:
> Is it the expected behaviour to have a success exit code when skip fails, even with the notice message "cannot skip..."?

That's the change that was put in a few years ago, yes:

http://git.savannah.gnu.org/cgit/coreutils.git/commit/?id=460ebb078ed7bf249af745d8725d6c37e7f15b57

As the changelog says, POSIX doesn't say what to do in this situation.
Traditional 'dd' (e.g., Solaris 11.1) exits with status 3 (!).

Message #11 received at 14897 <at> debbugs.gnu.org (full text, mbox):

From: Pádraig Brady <P <at> draigBrady.com>
To: Paul Eggert <eggert <at> cs.ucla.edu>
Cc: 14897 <at> debbugs.gnu.org, ffx <ffx-gdev <at> tfq.fr>
Subject: Re: bug#14897: dd skip exit code
Date: Thu, 18 Jul 2013 16:02:13 +0100

On 07/18/2013 03:40 PM, Paul Eggert wrote:
> On 07/18/2013 02:03 AM, ffx wrote:
>> Is it the expected behaviour to have a success exit code when skip fails, even with the notice message "cannot skip..."?
> 
> That's the change that was put in a few years ago, yes:
> 
> http://git.sv.gnu.org/cgit/coreutils.git/commit/?id=460ebb078

Right that commit added the new warning in this case

> As the changelog says, POSIX doesn't say what to do in this situation.

Right the code portion from the above commit explaining this is:

/* POSIX doesn't say what to do when dd detects it has been
   asked to skip past EOF, so I assume it's non-fatal.
   There are 3 reasons why there might be unskipped blocks/bytes:
     1. file is too small
     2. pipe has not enough data
     3. short reads   */

Hmm I wonder since this is informational, should it be suppressed with status=none
status=none was added recently and I'm leaning towards it
also supressing informational warnings like this?

> Traditional 'dd' (e.g., Solaris 11.1) exits with status 3 (!).

Interesting.

thanks,
Pádraig.

Message #14 received at 14897 <at> debbugs.gnu.org (full text, mbox):

From: Bernhard Voelker <mail <at> bernhard-voelker.de>
To: Pádraig Brady <P <at> draigBrady.com>
Cc: 14897 <at> debbugs.gnu.org, Paul Eggert <eggert <at> cs.ucla.edu>,
 ffx <ffx-gdev <at> tfq.fr>
Subject: Re: bug#14897: dd skip exit code
Date: Thu, 18 Jul 2013 17:16:07 +0200

On 07/18/2013 05:02 PM, Pádraig Brady wrote:
> /* POSIX doesn't say what to do when dd detects it has been
>    asked to skip past EOF, so I assume it's non-fatal.

OTOH, POSIX says:
(http://pubs.opengroup.org/onlinepubs/009695399/utilities/dd.html)

  [...] If some other error is detected, a diagnostic message shall
  be written and the copy operation shall be discontinued.

Hmm, whatever they mean with "some other error" - couldn't this
be such a seek-on-input failure?

Have a nice day,
Berny

Message #17 received at 14897 <at> debbugs.gnu.org (full text, mbox):

From: Paul Eggert <eggert <at> cs.ucla.edu>
To: Pádraig Brady <P <at> draigBrady.com>
Cc: 14897 <at> debbugs.gnu.org, ffx <ffx-gdev <at> tfq.fr>
Subject: Re: bug#14897: dd skip exit code
Date: Thu, 18 Jul 2013 08:50:31 -0700

On 07/18/13 08:02, Pádraig Brady wrote:
> Hmm I wonder since this is informational, should it be suppressed with status=none
> status=none was added recently and I'm leaning towards it
> also supressing informational warnings like this?

Yes, that makes sense.

It's not an error to seek past EOF, so it shouldn't be an error
and the Solaris behavior is (in my opinion) not that useful.
(At least, for regular files.  Not so sure about tapes.)

Message #20 received at 14897 <at> debbugs.gnu.org (full text, mbox):

From: Pádraig Brady <P <at> draigBrady.com>
To: Bernhard Voelker <mail <at> bernhard-voelker.de>
Cc: 14897 <at> debbugs.gnu.org, ffx <ffx-gdev <at> tfq.fr>
Subject: Re: bug#14897: dd skip exit code
Date: Thu, 18 Jul 2013 16:56:39 +0100

On 07/18/2013 04:16 PM, Bernhard Voelker wrote:
> On 07/18/2013 05:02 PM, Pádraig Brady wrote:
>> /* POSIX doesn't say what to do when dd detects it has been
>>    asked to skip past EOF, so I assume it's non-fatal.
> 
> OTOH, POSIX says:
> (http://pubs.opengroup.org/onlinepubs/009695399/utilities/dd.html)
> 
>   [...] If some other error is detected, a diagnostic message shall
>   be written and the copy operation shall be discontinued.
> 
> Hmm, whatever they mean with "some other error" - couldn't this
> be such a seek-on-input failure?

In the context of the whole section there,
I'm presuming they just mean conv=noerror is
only significant for input errors.

seeking past end of input couldn't be
considered an error IMHO.

thanks,
Pádraig.

Message #25 received at 14897-done <at> debbugs.gnu.org (full text, mbox):

From: Pádraig Brady <P <at> draigBrady.com>
To: Paul Eggert <eggert <at> cs.ucla.edu>
Cc: 14897-done <at> debbugs.gnu.org, ffx <ffx-gdev <at> tfq.fr>
Subject: Re: bug#14897: dd skip exit code
Date: Thu, 18 Jul 2013 18:52:20 +0100

[Message part 1 (text/plain, inline)]

On 07/18/2013 04:50 PM, Paul Eggert wrote:
> On 07/18/13 08:02, Pádraig Brady wrote:
>> Hmm I wonder since this is informational, should it be suppressed with status=none
>> status=none was added recently and I'm leaning towards it
>> also supressing informational warnings like this?
> 
> Yes, that makes sense.
> 
> It's not an error to seek past EOF, so it shouldn't be an error
> and the Solaris behavior is (in my opinion) not that useful.
> (At least, for regular files.  Not so sure about tapes.)

The attached patch allows one to set status=none to suppress all diagnostics.

thanks,
Pádraig.

[dd-suppress-warnings.patch (text/x-patch, attachment)]

Message #28 received at 14897-done <at> debbugs.gnu.org (full text, mbox):

From: Paul Eggert <eggert <at> cs.ucla.edu>
To: Pádraig Brady <P <at> draigBrady.com>
Cc: 14897-done <at> debbugs.gnu.org, ffx <ffx-gdev <at> tfq.fr>
Subject: Re: bug#14897: dd skip exit code
Date: Thu, 18 Jul 2013 11:25:12 -0700

Thanks.

Message #31 received at 14897 <at> debbugs.gnu.org (full text, mbox):

From: Bernhard Voelker <mail <at> bernhard-voelker.de>
To: 14897 <at> debbugs.gnu.org, P <at> draigBrady.com, ffx-gdev <at> tfq.fr
Subject: Re: bug#14897: dd skip exit code
Date: Thu, 18 Jul 2013 23:43:15 +0200

On 07/18/2013 07:52 PM, Pádraig Brady wrote:

> * NEWS: Mention the change in behavior

s/$/./

> @@ -2010,7 +2013,9 @@ dd_copy (void)
>  
>        if (nread < 0)
>          {
> -          error (0, errno, _("error reading %s"), quote (input_file));
> +          if (!(conversions_mask & C_NOERROR) || !(status_flags & STATUS_NONE))
> +            error (0, errno, _("error reading %s"), quote (input_file));
> +
>            if (conversions_mask & C_NOERROR)
>              {
>                print_stats ();

This doesn't look like a non-fatal error to me.  Are you sure
we really want to suppress this with status=none?

Reading the Texinfo manual, I think this change in behavior should
also be documented:

          Do not print any informational messages to stderr.  Error
          messages are output as normal.

It probably should say something like "... any informational nor
warning messages ...".

Otherwise +1.

Have a nice day,
Berny

Message #34 received at 14897 <at> debbugs.gnu.org (full text, mbox):

From: Pádraig Brady <P <at> draigBrady.com>
To: Bernhard Voelker <mail <at> bernhard-voelker.de>
Cc: 14897 <at> debbugs.gnu.org, ffx-gdev <at> tfq.fr
Subject: Re: bug#14897: dd skip exit code
Date: Thu, 18 Jul 2013 23:14:44 +0100

On 07/18/2013 10:43 PM, Bernhard Voelker wrote:
> 
> On 07/18/2013 07:52 PM, Pádraig Brady wrote:
> 
>> * NEWS: Mention the change in behavior
> 
> s/$/./
> 
>> @@ -2010,7 +2013,9 @@ dd_copy (void)
>>  
>>        if (nread < 0)
>>          {
>> -          error (0, errno, _("error reading %s"), quote (input_file));
>> +          if (!(conversions_mask & C_NOERROR) || !(status_flags & STATUS_NONE))
>> +            error (0, errno, _("error reading %s"), quote (input_file));
>> +
>>            if (conversions_mask & C_NOERROR)
>>              {
>>                print_stats ();
> 
> This doesn't look like a non-fatal error to me.  Are you sure
> we really want to suppress this with status=none?

I wasn't sure about that one since POSIX says you _shall_ output
a diagnostic in that case.  But of course status=none is an extension,
so POSIX doesn't apply so.  Also with conv=noerror this is just a warning,
so for consistency it seems right to treat like all other warnings.

> Reading the Texinfo manual, I think this change in behavior should
> also be documented:
> 
>           Do not print any informational messages to stderr.  Error
>           messages are output as normal.
> 
> It probably should say something like "... any informational nor
> warning messages ...".

I will amend like that and push.

thanks for the review!
Pádraig.

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.