[Cc Vivek for a question inline] On Mon, Oct 8, 2018 at 6:25 AM, Amir Goldstein <amir73il@xxxxxxxxx> wrote: > In an attempt to dedup ~100 LOC, we ended up creating a tangled call > chain, whose branches merge and diverge in several points according to > the immutable c->tmpfile copy up mode. > > This call chain was hard to analyse for locking correctness because the > locking requirements for the c->tmpfile flow were very different from > the locking requirements for the !c->tmpfile flow (i.e. directory vs. > regulare file copy up). > > Split the copy up helpers of the c->tmpfile flow from those of the > !c->tmpfile (i.e. workdir) flow and remove the c->tmpfile mode from > copy up context. > > Suggested-by: Al Viro <viro@xxxxxxxxxxxxxxxxxx> > Signed-off-by: Amir Goldstein <amir73il@xxxxxxxxx> > --- > > Miklos, > > I've managed to mess up error handling in 3 out of 6 helpers in v1. > Better take a close look that I did not miss more... > > Thanks, > Amir. > > fs/overlayfs/copy_up.c | 155 ++++++++++++++++++++++++++++++----------- > 1 file changed, 114 insertions(+), 41 deletions(-) > > diff --git a/fs/overlayfs/copy_up.c b/fs/overlayfs/copy_up.c > index 1cc797a08a5b..78a5049f7a5a 100644 > --- a/fs/overlayfs/copy_up.c > +++ b/fs/overlayfs/copy_up.c > @@ -395,7 +395,6 @@ struct ovl_copy_up_ctx { > struct dentry *destdir; > struct qstr destname; > struct dentry *workdir; > - bool tmpfile; > bool origin; > bool indexed; > bool metacopy; > @@ -440,8 +439,14 @@ static int ovl_link_up(struct ovl_copy_up_ctx *c) > return err; > } > > -static int ovl_install_temp(struct ovl_copy_up_ctx *c, struct dentry *temp, > - struct dentry **newdentry) > +/* > + * Move temp file from workdir into place on upper dir. > + * Used when copying up directories and when upper fs doesn't support O_TMPFILE. > + * > + * Caller must hold ovl_lock_rename_workdir(). > + */ > +static int ovl_rename_temp(struct ovl_copy_up_ctx *c, struct dentry *temp, > + struct dentry **newdentry) > { > int err; > struct dentry *upper; > @@ -451,19 +456,40 @@ static int ovl_install_temp(struct ovl_copy_up_ctx *c, struct dentry *temp, > if (IS_ERR(upper)) > return PTR_ERR(upper); > > - if (c->tmpfile) > - err = ovl_do_link(temp, udir, upper); > - else > - err = ovl_do_rename(d_inode(c->workdir), temp, udir, upper, 0); > + err = ovl_do_rename(d_inode(c->workdir), temp, udir, upper, 0); > + if (!err) > + *newdentry = dget(temp); > + dput(upper); > > + return err; > +} > + > +/* Link O_TMPFILE into place on upper dir */ > +static int ovl_link_tmpfile(struct ovl_copy_up_ctx *c, struct dentry *temp, > + struct dentry **newdentry) > +{ > + int err; > + struct dentry *upper = NULL; > + struct inode *udir = d_inode(c->destdir); > + > + inode_lock_nested(udir, I_MUTEX_PARENT); > + > + upper = lookup_one_len(c->destname.name, c->destdir, c->destname.len); > + err = PTR_ERR(upper); > + if (IS_ERR(upper)) > + goto out; > + > + err = ovl_do_link(temp, udir, upper); > if (!err) > - *newdentry = dget(c->tmpfile ? upper : temp); > + *newdentry = dget(upper); > dput(upper); > > +out: > + inode_unlock(udir); > return err; > } > > -static struct dentry *ovl_get_tmpfile(struct ovl_copy_up_ctx *c) > +static struct dentry *ovl_get_workdir_temp(struct ovl_copy_up_ctx *c) > { > int err; > struct dentry *temp; > @@ -484,10 +510,32 @@ static struct dentry *ovl_get_tmpfile(struct ovl_copy_up_ctx *c) > if (new_creds) > old_creds = override_creds(new_creds); > > - if (c->tmpfile) > - temp = ovl_do_tmpfile(c->workdir, c->stat.mode); > - else > - temp = ovl_create_temp(c->workdir, &cattr); > + temp = ovl_create_temp(c->workdir, &cattr); > +out: > + if (new_creds) { > + revert_creds(old_creds); Not new in this patch, but it looks like this will Oops if old_creds is NULL, which happens if security_inode_copy_up() returns an error. Trivial to fix, but I'm not sure we need the put_cred(new_creds) in the failed security_inode_copy_up() case. Vivek, do you remember the reason for this error cleanup? > + put_cred(new_creds); > + } > + > + return temp; > +} > + > +static struct dentry *ovl_get_tmpfile(struct ovl_copy_up_ctx *c) > +{ > + int err; > + struct dentry *temp; > + const struct cred *old_creds = NULL; > + struct cred *new_creds = NULL; > + > + err = security_inode_copy_up(c->dentry, &new_creds); > + temp = ERR_PTR(err); > + if (err < 0) > + goto out; > + > + if (new_creds) > + old_creds = override_creds(new_creds); > + > + temp = ovl_do_tmpfile(c->destdir, c->stat.mode); This changes c->workdir to c->destdir. Shouldn't normally matter, but... I changed this back, and we can discuss the merits of this change in the context of a separate patch. > out: > if (new_creds) { > revert_creds(old_creds); > @@ -548,37 +596,39 @@ static int ovl_copy_up_inode(struct ovl_copy_up_ctx *c, struct dentry *temp) > return err; > } > > -static int ovl_copy_up_locked(struct ovl_copy_up_ctx *c) > +/* > + * Copyup using workdir to prepare temp file. > + * Used when copying up directories and when upper fs doesn't support O_TMPFILE. > + */ > +static int ovl_copy_up_workdir(struct ovl_copy_up_ctx *c) > { > - struct inode *udir = c->destdir->d_inode; > struct inode *inode; > struct dentry *newdentry = NULL; > struct dentry *temp; > int err; > > - temp = ovl_get_tmpfile(c); > + err = ovl_lock_rename_workdir(c->workdir, c->destdir); > + if (err) > + return err; > + > + temp = ovl_get_workdir_temp(c); > + err = PTR_ERR(temp); > if (IS_ERR(temp)) > - return PTR_ERR(temp); > + goto out; > > err = ovl_copy_up_inode(c, temp); > if (err) > - goto out; > + goto out_cleanup; > > if (S_ISDIR(c->stat.mode) && c->indexed) { > err = ovl_create_index(c->dentry, c->lowerpath.dentry, temp); > if (err) > - goto out; > + goto out_cleanup; > } > > - if (c->tmpfile) { > - inode_lock_nested(udir, I_MUTEX_PARENT); > - err = ovl_install_temp(c, temp, &newdentry); > - inode_unlock(udir); > - } else { > - err = ovl_install_temp(c, temp, &newdentry); > - } > + err = ovl_rename_temp(c, temp, &newdentry); > if (err) > - goto out; > + goto out_cleanup; > > if (!c->metacopy) > ovl_set_upperdata(d_inode(c->dentry)); > @@ -587,10 +637,41 @@ static int ovl_copy_up_locked(struct ovl_copy_up_ctx *c) > if (S_ISDIR(inode->i_mode)) > ovl_set_flag(OVL_WHITEOUTS, inode); > > -out: > - if (err && !c->tmpfile) > +out_cleanup: > + if (err) > ovl_cleanup(d_inode(c->workdir), temp); This is a bit awkward: ovl_cleanup() will be called IFF we'd jumped to out_cleanup. So I moved this error cleanup till after the return. > dput(temp); > +out: > + unlock_rename(c->workdir, c->destdir); > + return err; > + > +} > + > +/* Copyup using O_TMPFILE which does not require cross dir locking */ > +static int ovl_copy_up_tmpfile(struct ovl_copy_up_ctx *c) > +{ > + struct dentry *newdentry = NULL; > + struct dentry *temp; > + int err; > + > + temp = ovl_get_tmpfile(c); > + if (IS_ERR(temp)) > + return PTR_ERR(temp); > + > + err = ovl_copy_up_inode(c, temp); > + if (err) > + goto out; > + > + err = ovl_link_tmpfile(c, temp, &newdentry); > + if (err) > + goto out; > + > + if (!c->metacopy) > + ovl_set_upperdata(d_inode(c->dentry)); > + ovl_inode_update(d_inode(c->dentry), newdentry); > + > +out: > + dput(temp); > return err; > > } > @@ -646,18 +727,10 @@ static int ovl_do_copy_up(struct ovl_copy_up_ctx *c) > } > > /* Should we copyup with O_TMPFILE or with workdir? */ > - if (S_ISREG(c->stat.mode) && ofs->tmpfile) { > - c->tmpfile = true; > - err = ovl_copy_up_locked(c); > - } else { > - err = ovl_lock_rename_workdir(c->workdir, c->destdir); > - if (!err) { > - err = ovl_copy_up_locked(c); > - unlock_rename(c->workdir, c->destdir); > - } > - } > - > - > + if (S_ISREG(c->stat.mode) && ofs->tmpfile) > + err = ovl_copy_up_tmpfile(c); > + else > + err = ovl_copy_up_workdir(c); > if (err) > goto out; > > -- > 2.17.1 >