| Message ID | 20150227235100.GA12964@devel.8.8.4.4 (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 2015/2/28 7:51, Daeseok Youn wrote: > There is no error handle when ocfs2_journal_access_di() is failed. > And also it doesn't need to call ocfs2_dx_dir_insert() when > ocfs2_journal_access_db() is failed. > > Signed-off-by: Daeseok Youn <daeseok.youn@gmail.com> > --- > RESEND: this patch rebased by 1/4 > > fs/ocfs2/dir.c | 8 ++++---- > 1 files changed, 4 insertions(+), 4 deletions(-) > > diff --git a/fs/ocfs2/dir.c b/fs/ocfs2/dir.c > index c1ab24f..c63f2b6 100644 > --- a/fs/ocfs2/dir.c > +++ b/fs/ocfs2/dir.c > @@ -1703,15 +1703,15 @@ int __ocfs2_add_entry(handle_t *handle, > insert_bh, > OCFS2_JOURNAL_ACCESS_WRITE); > > - if (ocfs2_dir_indexed(dir)) { > + if (!retval && ocfs2_dir_indexed(dir)) > retval = ocfs2_dx_dir_insert(dir, > handle, > lookup); > - if (retval) > - goto bail; > - } > } > > + if (retval) > + goto bail; > + > /* By now the buffer is marked for journaling */ > offset += le16_to_cpu(de->rec_len); > if (le64_to_cpu(de->inode)) { > I agree with you that we should do the corresponding check. But I don't think we need remove the mlog as described in my previous mail.
OK. I will send this patch again. This patch will be based on https://lkml.org/lkml/2015/2/27/655 thanks regards, Daeseok Youn 2015-03-19 20:00 GMT+09:00 Joseph Qi <joseph.qi@huawei.com>: > On 2015/2/28 7:51, Daeseok Youn wrote: >> There is no error handle when ocfs2_journal_access_di() is failed. >> And also it doesn't need to call ocfs2_dx_dir_insert() when >> ocfs2_journal_access_db() is failed. >> >> Signed-off-by: Daeseok Youn <daeseok.youn@gmail.com> >> --- >> RESEND: this patch rebased by 1/4 >> >> fs/ocfs2/dir.c | 8 ++++---- >> 1 files changed, 4 insertions(+), 4 deletions(-) >> >> diff --git a/fs/ocfs2/dir.c b/fs/ocfs2/dir.c >> index c1ab24f..c63f2b6 100644 >> --- a/fs/ocfs2/dir.c >> +++ b/fs/ocfs2/dir.c >> @@ -1703,15 +1703,15 @@ int __ocfs2_add_entry(handle_t *handle, >> insert_bh, >> OCFS2_JOURNAL_ACCESS_WRITE); >> >> - if (ocfs2_dir_indexed(dir)) { >> + if (!retval && ocfs2_dir_indexed(dir)) >> retval = ocfs2_dx_dir_insert(dir, >> handle, >> lookup); >> - if (retval) >> - goto bail; >> - } >> } >> >> + if (retval) >> + goto bail; >> + >> /* By now the buffer is marked for journaling */ >> offset += le16_to_cpu(de->rec_len); >> if (le64_to_cpu(de->inode)) { >> > I agree with you that we should do the corresponding check. > But I don't think we need remove the mlog as described in my previous > mail. >
diff --git a/fs/ocfs2/dir.c b/fs/ocfs2/dir.c index c1ab24f..c63f2b6 100644 --- a/fs/ocfs2/dir.c +++ b/fs/ocfs2/dir.c @@ -1703,15 +1703,15 @@ int __ocfs2_add_entry(handle_t *handle, insert_bh, OCFS2_JOURNAL_ACCESS_WRITE); - if (ocfs2_dir_indexed(dir)) { + if (!retval && ocfs2_dir_indexed(dir)) retval = ocfs2_dx_dir_insert(dir, handle, lookup); - if (retval) - goto bail; - } } + if (retval) + goto bail; + /* By now the buffer is marked for journaling */ offset += le16_to_cpu(de->rec_len); if (le64_to_cpu(de->inode)) {
There is no error handle when ocfs2_journal_access_di() is failed. And also it doesn't need to call ocfs2_dx_dir_insert() when ocfs2_journal_access_db() is failed. Signed-off-by: Daeseok Youn <daeseok.youn@gmail.com> --- RESEND: this patch rebased by 1/4 fs/ocfs2/dir.c | 8 ++++---- 1 files changed, 4 insertions(+), 4 deletions(-)