PƔdraig Brady <P@draigBrady.com> writes:
> On 16/03/2026 21:43, Bruno Haible via GNU coreutils General Discussion wrote:
>> The CI reports a test failure of a new test on NetBSD 10.0:
>>
>> FAIL: tests/dd/partial-write
>> ============================
>>
>> dd: error writing 'f': File too large
>> 1+0 records in
>> 0+0 records out
>> 0 bytes copied, 0.000367544 s, 0.0 kB/s
>> FAIL tests/dd/partial-write.sh (exit status: 1)
>
> Interesting, there are 0+0 records out,
> so no partial data written at all (I presume).
> We could probably handle that with the following
> (I don't have access to netbsd (on cfarm) to test).
>
> cheers,
> Padraig
>
> diff --git a/tests/dd/partial-write.sh b/tests/dd/partial-write.shindex
> 2c3019491..1f4ab7895 100755
> --- a/tests/dd/partial-write.sh
> +++ b/tests/dd/partial-write.sh
> @@ -27,6 +27,7 @@ print_ver_ dd
> ret=$?
>
> if test $ret = 1; then
> + test -s f || skip 'The system disallowed all writes'
> grep -F '+1 records out' err || { cat err; fail=1; }
> elif test $ret = 0; then
> skip_ 'The system did not limit the file fize'
I could reproduce the failure on a NetBSD 10.1 virtual machine. After
applying this patch with the s/skip/skip_/ change, I confirm the test is
skipped.
Collin