[parted-devel] [PATCH] libparted: remove device mapper node for last partition

[Jim Meyering]

Phillip Susi wrote:
> On 8/8/2012 11:23 AM, Jim Meyering wrote:
>> Sorry about the delay. I see that the most recent patch had a
>> misleading comment for which you proposed wording to correct it,
>> but no amended patch. Would you please resend a complete, corrected
>> patch?
>
> That would be the fix for optimal IO alignment.  I'll update the
> comment and resubmit the patch for that as well, but the series I was
> referring to was "dmraid fixes" on Jan 8, specifically "refactor
> device-mapper partition sync code".
>
> Looking back over that thread, the unresolved issue appears to be with
> t2310-dos-extended-2-sector-min-offset.sh.  I discovered that the test
> is fundamentally broken because it appears to be supposed to validate
> that parted refuses to create an extended partition without the 2
> sector gap for LILO, but in fact, parted happily creates such a
> partition, but emits an error when it tries to sync the kernel
> partition table to what it just wrote to disk, and it is that error
> that this test is looking for.
>
> I had generated a patch to fix parted to avoid the error syncing the
> kernel, which would render t2310-dos-extended-2-sector-min-offset.sh
> obsolete, but asked whether you wanted to go that route, or to
> actually change parted to refuse to create such a partition.  I see no
> reason why parted should refuse to create such a partition since the
> kernel is happy with it.  Unless you object, I think I will rebase and
> resubmit the whole series with the patch fixing the syncing and
> removing t2310-dos-extended-2-sector-min-offset.sh.

Thank you.
Yes, that sounds like the right approach.