From mboxrd@z Thu Jan 1 00:00:00 1970 From: 'Tyler Hicks' Subject: Re: [PATCH] eCryptfs: Quota check incorrectly ignored Date: Fri, 10 Feb 2012 13:41:42 -0600 Message-ID: <20120210194142.GB15606@boyd> References: <1328787572-2030-1-git-send-email-liwang@nudt.edu.cn> <528868922.02520@eyou.net> <000001cce802$709bf770$51d3e650$@edu.cn> <20120210193148.GA15606@boyd> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha512; protocol="application/pgp-signature"; boundary="UHN/qo2QbUvPLonB" Cc: 'Jan Kara' , ecryptfs@vger.kernel.org, linux-kernel@vger.kernel.org, 'Yong Peng' , viro@zeniv.linux.org.uk, linux-fsdevel@vger.kernel.org, akpm@linux-foundation.org, Thieu Le To: Li Wang Return-path: Content-Disposition: inline In-Reply-To: <20120210193148.GA15606@boyd> Sender: ecryptfs-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org --UHN/qo2QbUvPLonB Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On 2012-02-10 13:31:49, 'Tyler Hicks' wrote: > On 2012-02-10 22:44:05, Li Wang wrote: > > > -----Original Message----- > > > From: Jan Kara [mailto:jack@suse.cz] > > > Sent: Friday, February 10, 2012 6:32 PM > > > To: Li Wang > > > Cc: Tyler Hicks; ecryptfs@vger.kernel.org; linux-kernel@vger.kernel.o= rg; > > Jan > > > Kara; Yong Peng > > > Subject: Re: [PATCH] eCryptfs: Quota check incorrectly ignored > > >=20 > > > On Thu 09-02-12 19:39:32, Li Wang wrote: > > > > eCryptfs recently modified the write path to perform encryption > > > > and write down in ecryptfs_writepage(). This function invokes > > vfs_write() > > > > to write down the encrypted data to lower page cache. vfs_write() w= ill > > > > first make sure this write will not exceed the quota limit for the = owner > > > > of the file being written into, if quota is supported by > > > > the underlying file system, and it is turned on. Normally, it > > accomplishs > > > > this job by calling check_idq()/check_bdq() (fs/quota/dquot.c). When > > system > > > > dirty ratio is not high, ecryptfs_writepage() is normally invoked b= y the > > > > write back kernel thread, who has the capability CAP_SYS_RESOURCE, > > > > this priority will let check_idq()/check_bdq() directly bypass the = quota > > > > check. This sometimes makes data safely written into the disk in sp= ite > > of > > > > exceeding the quota limit. > > > > > > > > This patch temporarily removes the CAP_SYS_RESOURCE capability from= the > > > kernel > > > > thread before invoking vfs_write_lower(), to let it undergo quota c= heck > > by > > > > the lower file system, if necessary. After that, reassign the > > capability. > > > Hmm, but then the error will just be thrown away by the flusher thr= ead > > > and the application never learns about it? That doesn't sound like a = good > > > solution. > > >=20 > > > Honza > >=20 > > Yes. we are aware of that, but we have not found better solution, since= it > > seems=20 > > VFS does not supply a file system independent interface to check quota = early > > and only > > (fix us if we are wrong), The file systems just perform the quota check= in > > their own way,=20 > > the routine is wrapped deeply inside the file system specific code path. > > Maybe we could > > copy some codes from _dquot_alloc_space() & dquot_alloc_inode > > (fs/quota/dquot.c) to=20 > > perform quota check on lower inode ourselves, but that is ugly, and we = are > > not sure if it works > > for any file system or not..., On the other side, due to the existence = of > > write buffer, and io schedule, > > the successful write call does not gurantee the data will be written in= to > > disk, in terms of that, > > we do not change much of the semantic of write call.=20 >=20 > Maybe I should just revert 57db4e8d73ef2b5e94a3f412108dff2576670a8a > (and friends) and go back to the write-through cache model for eCryptfs. > The write-back model has some nice performance improvements, but we keep > running into little issues like this. For write-through to work > reliably, we're going to need some support from the VFS that isn't there > yet. Sorry, that should be "For write-back to work reliably, we're going to need some support from the VFS that isn't there yet." Tyler >=20 > Handling ENOSPC correctly is another area that is lacking after the > switch to write-back. Currently, an application will continue to write > out data without having any indication that the disk is full. Thieu > proposed a fix, but it was never merged: >=20 > http://article.gmane.org/gmane.linux.kernel/1209265 >=20 > So, what do folks think about going back to a write-through cache until > we have the ability to alert the application of these error conditions? >=20 > Tyler >=20 > > =20 > > >=20 > > >=20 > > > > > > > > Signed-off-by: Li Wang > > > > Singed-off-by: Yong Peng > > > > --- > > > > > > > > To repeat this bug, > > > > mount -o usrquota /dev/sda3 /tmp > > > > cd /tmp > > > > edquota -u foo // set the disk quota limit for user foo be m1 bytes > > > > quotaon -a > > > > mount -t ecryptfs cipher plain > > > > > > > > login the system as user foo > > > > cd /tmp/plain > > > > execute the following simple program > > > > > > > > int main() > > > > { > > > > char buf[m2]; // m2>m1 > > > > FILE *f =3D fopen("dummy", "w"); > > > > fwrite(buf, 1, m2, f); > > > > sleep(60); // let the kernel thread do the write back job > > > > fclose(f); > > > > return 0; > > > > } > > > > > > > > This program can write as much of data as it wants, provided sleep > > > enough long time before closing the file. > > > > > > > > fs/ecryptfs/crypto.c | 27 +++++++++++++++++++++++++++ > > > > 1 files changed, 27 insertions(+), 0 deletions(-) > > > > > > > > diff --git a/fs/ecryptfs/crypto.c b/fs/ecryptfs/crypto.c > > > > index 63ab245..2c1da29 100644 > > > > --- a/fs/ecryptfs/crypto.c > > > > +++ b/fs/ecryptfs/crypto.c > > > > @@ -457,6 +457,7 @@ int ecryptfs_encrypt_page(struct page *page) > > > > struct page *enc_extent_page =3D NULL; > > > > loff_t extent_offset; > > > > int rc =3D 0; > > > > + struct cred *cred; > > > > > > > > ecryptfs_inode =3D page->mapping->host; > > > > crypt_stat =3D > > > > @@ -487,8 +488,34 @@ int ecryptfs_encrypt_page(struct page *page) > > > > * (PAGE_CACHE_SIZE > > > > / crypt_stat->extent_size)) > > > > + extent_offset), crypt_stat); > > > > + if (current->flags & PF_KTHREAD) { > > > > + /* > > > > + * Temporarily remove the > > > CAP_SYS_RESOURCE capability > > > > + * from the write back kernel thread to le= t it > > > undergo > > > > + * quota check by the lower file system > > > > + */ > > > > + cred =3D prepare_creds(); > > > > + if (unlikely(!cred)) { > > > > + rc =3D -ENOMEM; > > > > + goto out; > > > > + } > > > > + cap_lower(cred->cap_effective, CAP_SYS_RESOURCE); > > > > + commit_creds(cred); > > > > + } > > > > rc =3D ecryptfs_write_lower(ecryptfs_inode, enc_extent_virt, > > > > offset, crypt_stat->extent_size); > > > > + if (current->flags & PF_KTHREAD) { > > > > + /* > > > > + * Reassign the CAP_SYS_RESOURCE > > > capability > > > > + */ > > > > + cred =3D prepare_creds(); > > > > + if (unlikely(!cred)) { > > > > + rc =3D -ENOMEM; > > > > + goto out; > > > > + } > > > > + cap_raise(cred->cap_effective, CAP_SYS_RESOURCE); > > > > + commit_creds(cred); > > > > + } > > > > if (rc < 0) { > > > > ecryptfs_printk(KERN_ERR, "Error attempting " > > > > "to write lower page; rc =3D [%d]" > > > > -- > > > > 1.7.6.5 > > > -- > > > Jan Kara > > > SUSE Labs, CR > >=20 --UHN/qo2QbUvPLonB Content-Type: application/pgp-signature; name="signature.asc" Content-Description: Digital signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux) iQIcBAEBCgAGBQJPNXL2AAoJENaSAD2qAscKM5gP/0xgOUD6+J8kgnT715O6OAQT 3ZHpEBqt+hDFpX+ZZKkDT415epqznc4MAhjWJlB8fnD1RDLC5WIksK+gzF4eab9s cfC6PC+v+UFFywGg5kKuD5VdAvXdGCkF+DlatpWj4+3r34G5C4IxDAy2K/OVRBAQ vYVgdBoNQdbUqRVG2B54C6cVmp8XtsD4NOM+89jObFEg3NrzBYqGtTxfuAXgJAkw d1tcYbDw3i862uIdqxeBSrv4KnhL2WMO7Qi95oX3tGvYjuGnMppvT9sm4oBL1Gqw 9t16oOgINg3wI6vpr5M3S6FtqLcu6omXlqq+5A/7rLD46DWqLxIZPyDFZNxhDtWp cbdOdnoKj2KyU072L9SMfhSiwZv93Dh6zwSnP3ywav1jPC/fKP8DbH/CvFJ7o6C5 cXGdqcXCTZfqXyaDt1EO99UbxRqZscpioKNfkxTPk490aCe53gjCSnqRBmPtGlRT 4G4oo1b7oE4XcXfDZf8PRJvnpPob4Yr0aBcjTev19W6kmK5H31oU0V+jDKHvmCyz kOXOy/X0I3okVjHLr2mpEQjYxd+kgUi/g2bAIkM5DYcCuJXFJ/RhI2as2PeDf90l d4mERyd1lEOof7r3J0yPpB5KnNInQH3EmpnOVBMZ/8AdFP6tRI6MgDW+IAd/KtYl x7vzhLtj6LNc/765UJg/ =isQ4 -----END PGP SIGNATURE----- --UHN/qo2QbUvPLonB--