| Message ID | 1428097580-25352-2-git-send-email-tariq.x.saeed@oracle.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
Hi Tariq, On 04/04/2015 05:46 AM, Tariq Saeed wrote: > Orabug: 20189959 > > This bug in mainline code is pointed out by Mark Fasheh. When ocfs2_iop_set_acl > and ocfs2_iop_ge_acl are entered from VFS layer, inode lock is not held. This > seems to be regression from older kernels. The patch is to fix that. > > Signed-off-by: Tariq Saeed <tariq.x.saeed@oracle.com> > --- > fs/ocfs2/acl.c | 28 +++++++++++++++++++++------- > 1 files changed, 21 insertions(+), 7 deletions(-) > > diff --git a/fs/ocfs2/acl.c b/fs/ocfs2/acl.c > index 7e8282d..d7b5542 100644 > --- a/fs/ocfs2/acl.c > +++ b/fs/ocfs2/acl.c > @@ -286,7 +286,19 @@ int ocfs2_set_acl(handle_t *handle, > > int ocfs2_iop_set_acl(struct inode *inode, struct posix_acl *acl, int type) > { > - return ocfs2_set_acl(NULL, inode, NULL, type, acl, NULL, NULL); > + struct buffer_head *bh = NULL; > + int status = 0; > + > + status = ocfs2_inode_lock(inode, &bh, 1); > + if (status < 0) { > + if (status != -ENOENT) > + mlog_errno(status); > + return status; > + } > + status = ocfs2_set_acl(NULL, inode, bh, type, acl, NULL, NULL); > + ocfs2_inode_unlock(inode, 1); > + brelse(bh); > + return status; > } > > struct posix_acl *ocfs2_iop_get_acl(struct inode *inode, int type) > @@ -294,19 +306,21 @@ struct posix_acl *ocfs2_iop_get_acl(struct inode *inode, int type) > struct ocfs2_super *osb; > struct buffer_head *di_bh = NULL; > struct posix_acl *acl; > - int ret = -EAGAIN; > + int ret; > > osb = OCFS2_SB(inode->i_sb); > if (!(osb->s_mount_opt & OCFS2_MOUNT_POSIX_ACL)) > return NULL; > - > - ret = ocfs2_read_inode_block(inode, &di_bh); > - if (ret < 0) > - return ERR_PTR(ret); > + ret = ocfs2_inode_lock(inode, &di_bh, 0); > + if (ret < 0) { > + mlog_errno(ret); > + acl = ERR_PTR(ret); > + return acl; I think "return ERR_PTR(ret);" is more simple here. Also why you check "status != -ENOENT" in ocfs2_iop_set_acl() but not here? Thanks, Junxiao. > + } > > acl = ocfs2_get_acl_nolock(inode, type, di_bh); > > + ocfs2_inode_unlock(inode, 0); > brelse(di_bh); > - > return acl; > } >
On 04/16/2015 01:48 AM, Junxiao Bi wrote: > I think "return ERR_PTR(ret);" is more simple here. True. I am guilty of just copy-paste from an older kernel. > Also why you check "status != -ENOENT" in ocfs2_iop_set_acl() but not here Same reason -- copy-and-paste from an older kernel. Having said that, I don't think it matters one way or the other. Does it? I have no objection to applying your suggestions and re-submitting the patch. Mark, what is the protocol for that. Do I need to mention that this is a re-sumit? Thanks, -Tariq Saeed
On Thu, Apr 16, 2015 at 11:51:41AM -0700, Tariq Saeed wrote: > I have no objection to applying your suggestions and re-submitting the > patch. Mark, what is the protocol for that. Do I need to mention > that this is a re-sumit? If you're re-submitting something that hasn't gone upstream yet it's sufficient to add a 'V2' (or whatever you like) to the subject line. Also, a change log in those cases is nice. Thanks, --Mark -- Mark Fasheh
On 04/16/2015 02:15 PM, Mark Fasheh wrote:
> change log in those cases is nice.
Sorry, I don't understand what you mean by 'change log'.
I have not seen any message about this patch's -v1 into being
queued for main stream (unless the message is in transit -- very
unlikely!). So, basically, how do I request to discard the first version ...
or is that request implicit in submitting -v2?
Thanks
-Tariq
On Thu, Apr 16, 2015 at 02:23:52PM -0700, Tariq Saeed wrote: > > On 04/16/2015 02:15 PM, Mark Fasheh wrote: > > change log in those cases is nice. > Sorry, I don't understand what you mean by 'change log'. By changelog, I mean "what changed from the previous version of this patch". For example: Changes from v1 to v2 of this patch: - whitespace cleanups - fixed a bug in foobar() ... > I have not seen any message about this patch's -v1 into being > queued for main stream (unless the message is in transit -- very > unlikely!). So, basically, how do I request to discard the first version ... > or is that request implicit in submitting -v2? Yeah that's implicit in the -v2 being there. --Mark -- Mark Fasheh
diff --git a/fs/ocfs2/acl.c b/fs/ocfs2/acl.c index 7e8282d..d7b5542 100644 --- a/fs/ocfs2/acl.c +++ b/fs/ocfs2/acl.c @@ -286,7 +286,19 @@ int ocfs2_set_acl(handle_t *handle, int ocfs2_iop_set_acl(struct inode *inode, struct posix_acl *acl, int type) { - return ocfs2_set_acl(NULL, inode, NULL, type, acl, NULL, NULL); + struct buffer_head *bh = NULL; + int status = 0; + + status = ocfs2_inode_lock(inode, &bh, 1); + if (status < 0) { + if (status != -ENOENT) + mlog_errno(status); + return status; + } + status = ocfs2_set_acl(NULL, inode, bh, type, acl, NULL, NULL); + ocfs2_inode_unlock(inode, 1); + brelse(bh); + return status; } struct posix_acl *ocfs2_iop_get_acl(struct inode *inode, int type) @@ -294,19 +306,21 @@ struct posix_acl *ocfs2_iop_get_acl(struct inode *inode, int type) struct ocfs2_super *osb; struct buffer_head *di_bh = NULL; struct posix_acl *acl; - int ret = -EAGAIN; + int ret; osb = OCFS2_SB(inode->i_sb); if (!(osb->s_mount_opt & OCFS2_MOUNT_POSIX_ACL)) return NULL; - - ret = ocfs2_read_inode_block(inode, &di_bh); - if (ret < 0) - return ERR_PTR(ret); + ret = ocfs2_inode_lock(inode, &di_bh, 0); + if (ret < 0) { + mlog_errno(ret); + acl = ERR_PTR(ret); + return acl; + } acl = ocfs2_get_acl_nolock(inode, type, di_bh); + ocfs2_inode_unlock(inode, 0); brelse(di_bh); - return acl; }
Orabug: 20189959 This bug in mainline code is pointed out by Mark Fasheh. When ocfs2_iop_set_acl and ocfs2_iop_ge_acl are entered from VFS layer, inode lock is not held. This seems to be regression from older kernels. The patch is to fix that. Signed-off-by: Tariq Saeed <tariq.x.saeed@oracle.com> --- fs/ocfs2/acl.c | 28 +++++++++++++++++++++------- 1 files changed, 21 insertions(+), 7 deletions(-)