Re: [PATCH 4/4] xfs: test upgrading filesystem to bigtime

From: Amir Goldstein <amir73il@gmail.com>
To: "Darrick J. Wong" <darrick.wong@oracle.com>
Cc: Eryu Guan <guaneryu@gmail.com>,
	linux-xfs <linux-xfs@vger.kernel.org>,
	fstests <fstests@vger.kernel.org>,
	Deepa Dinamani <deepa.kernel@gmail.com>
Subject: Re: [PATCH 4/4] xfs: test upgrading filesystem to bigtime
Date: Tue, 18 Aug 2020 21:40:26 +0300	[thread overview]
Message-ID: <CAOQ4uxgzMD_Zi90OwYiv2PH7rRs+W7MSYU3Zf0KVG1Kw-rQ-KA@mail.gmail.com> (raw)
In-Reply-To: <20200818182322.GX6096@magnolia>

> > Darrick,
> >
> > These tests look great, but I wonder.
> > generic/402 has more test coverage than above.
> > It tests several data points and it tests them with and without mount cycle.
>
> Um... these two tests exist to make sure that /upgrading/ works, whereas
> generic/402 tests whatever it finds on the formatted scratch filesystem.
>
> > With your current tests, bigtime will enjoy this test coverage only if
> > the entire
> > run is configured with custom XFS_MKFS_OPTIONS or when bigtime
> > becomes default for mkfs.
>
> I don't understand the line of reasoning.  Both tests format with
> specific mkfs options, or skip the test entirely if mkfs doesn't know
> what bigtime is.
>

I was referring to generic/402.

> > Do you think we should have a temporary clone of generic/402 for xfs which
> > enables bigtime for the time being?
>
> <shrug> I pushed most of my testing to the cloud, so I just spawn enough
> VMs so that one of them will test bigtime=0 and another does
> bigtime=1...
>

Fine by me. As long as there is test coverage for generic/402 with bigtime.

Thanks,
Amir.