From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from cantor2.suse.de ([195.135.220.15]:48259 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751533Ab3IRO31 (ORCPT ); Wed, 18 Sep 2013 10:29:27 -0400 Message-ID: <5239B8C2.4070107@suse.com> Date: Wed, 18 Sep 2013 10:29:22 -0400 From: Jeff Mahoney MIME-Version: 1.0 To: dsterba@suse.cz, linux-btrfs@vger.kernel.org, clmason@fusionio.com Subject: Re: [patch 1/9] btrfs: add ioctls to query/change feature bits online References: <20130916181910.799140428@suse.com> <20130916182017.776301648@suse.com> <20130918141459.GA6810@twin.jikos.cz> In-Reply-To: <20130918141459.GA6810@twin.jikos.cz> Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="Ml94uKLMeOe1nbgUTDLX7R0pSLtwHvro8" Sender: linux-btrfs-owner@vger.kernel.org List-ID: This is an OpenPGP/MIME signed message (RFC 4880 and 3156) --Ml94uKLMeOe1nbgUTDLX7R0pSLtwHvro8 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable On 9/18/13 10:14 AM, David Sterba wrote: > On Mon, Sep 16, 2013 at 02:19:11PM -0400, Jeff Mahoney wrote: >> There are some feature bits that require no offline setup and can >> be enabled online. I've only reviewed extended irefs, but there will >> probably be more. >=20 > Each feature bit being set online has to be treated separately regardin= g > safety of the data in flight. >=20 > You can set extended irefs online, there's only one place where the bit= > is checked (if the hadlink count overflows the leaf), it's ok to set th= e > bit anytime before or after and it will not break anything. >=20 > Skinny metadata bit is checked all over the place and I'm not sure it's= > ok to set it on after some portion of data has been processed in the > non-skinny mode. >=20 > 'default_subvol', lzo are ok, the rest depends on mkfs and can't be > modified online at all. >=20 > So, for skinny metadata it's most likely required to set the bit in a > delayed manner and schedule it eg. to post-commit time. Yep, that's why extended_iref is the only one I've enabled so far. Other features with harder requirements are going to need to be handled explicitly. > The sysfs and ioctl interface looks ok to me. Great, thanks. -Jeff --=20 Jeff Mahoney SUSE Labs --Ml94uKLMeOe1nbgUTDLX7R0pSLtwHvro8 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG/MacGPG2 v2.0.19 (Darwin) iQIcBAEBAgAGBQJSObjCAAoJEB57S2MheeWyXhcQAJndIMroY6S7ibrHDpxyThFC nrHUJNGaTSNChmcdjfFn2EszA5umPOdqAAucY2QozY9xy/4KFyrdFESd9T9PIqK2 D0EcdiHNz6OFjCDMH0npVMC1YwTLRQ5dxjvn+6YPGyww+tyfy6Xe0zLomo9ar5U6 Hyux6DvkVKtFn2ce/w6rtLFrVe7gHfJdTfCNJDT9EJjb7XO6DqUhZyWN7hELs5r1 NOG/TjgMW5XC+kQOYRYzn/GqD+XBtFe13lRTCByfH9r1DbN2v534XIJSHcp23tjl u/CzTECgLaAlkuHBMolYbHzoeVXNl1JYYSuYxrGQpXoylAGQSquXtmazBw/r+VlH m92pIrnDfywz0s+yMFajrKqVbJK2CFBqX9HrHR4H1dMUgum/uarZbYTU50gogHpJ /GL41NywA6iHFwFGx7HOkW8fGmBOlPRyRdLhK5rJfoV3JekUQeRVFQY9kcrjWXjM yGYzByXGiYsObER5QY5LbMfNFH2nTcIB3l6MCBh3PNOqWNHC/FL0rlG4+Xkd0ifE HZNdR3/uMrx5pIVe5/SGghH614V/F+CRgD1w4pFdPu+MivalrCiRFCzWlTXJC2Jb XwmSN2LcSqR4+GiIxt3Wt5b8JBcBFyW+lrDkAaFKm1xxzHW1JDxletMn2T5D+92C pb2uYjpJ2ghxRYkNraPo =y9WJ -----END PGP SIGNATURE----- --Ml94uKLMeOe1nbgUTDLX7R0pSLtwHvro8--