Message ID | 20210621152249.20901-1-colin.king@canonical.com |
---|---|
State | Rejected |
Headers | show |
Series | ubifs: Remove a redundant null check on pointer lp | expand |
在 2021/6/21 23:22, Colin King 写道: > From: Colin Ian King <colin.king@canonical.com> > > An earlier fix to replace an IS_ERR check on lp with a null check > on lp didn't remove a following null check on lp. The second null > check is redundant and can be removed. > > Addresses-Coverity: ("Logically dead code") > Fixes: c770cd5190ba ("ubifs: fix an IS_ERR() vs NULL check") > Signed-off-by: Colin Ian King <colin.king@canonical.com> > --- > fs/ubifs/gc.c | 2 -- > 1 file changed, 2 deletions(-) > > diff --git a/fs/ubifs/gc.c b/fs/ubifs/gc.c > index 7cc22d7317ea..465beea52176 100644 > --- a/fs/ubifs/gc.c > +++ b/fs/ubifs/gc.c > @@ -899,8 +899,6 @@ int ubifs_gc_start_commit(struct ubifs_info *c) > err = -ENOMEM; > goto out; > } Hi Colin, I just found out about it today thanks to your patch. Commit c770cd5190ba ("ubifs: fix an IS_ERR() vs NULL check") did import a new problem that ubifs_gc_start_commit() may return -ENOMEM while syncing fs. I guess ubifs_fast_find_frdi_idx() return NULL pointer is the termination condition in while-loop, which means we cannot get a freeable index LEB in ubifs_gc_start_commit(). > - if (!lp) > - break; > idx_gc = kmalloc(sizeof(struct ubifs_gced_idx_leb), GFP_NOFS); > if (!idx_gc) { > err = -ENOMEM; > BTW, the following modifications may be what you want? diff --git a/fs/ubifs/gc.c b/fs/ubifs/gc.c index 7cc22d7317ea..b1f276599b04 100644 --- a/fs/ubifs/gc.c +++ b/fs/ubifs/gc.c @@ -895,10 +895,6 @@ int ubifs_gc_start_commit(struct ubifs_info *c) /* Record index freeable LEBs for unmapping after commit */ while (1) { lp = ubifs_fast_find_frdi_idx(c); - if (!lp) { - err = -ENOMEM; - goto out; - } if (!lp) break; idx_gc = kmalloc(sizeof(struct ubifs_gced_idx_leb), GFP_NOFS);
On Tue, Jun 22, 2021 at 10:38:52AM +0800, Zhihao Cheng wrote: > 在 2021/6/21 23:22, Colin King 写道: > > From: Colin Ian King <colin.king@canonical.com> > > > > An earlier fix to replace an IS_ERR check on lp with a null check > > on lp didn't remove a following null check on lp. The second null > > check is redundant and can be removed. > > > > Addresses-Coverity: ("Logically dead code") > > Fixes: c770cd5190ba ("ubifs: fix an IS_ERR() vs NULL check") > > Signed-off-by: Colin Ian King <colin.king@canonical.com> > > --- > > fs/ubifs/gc.c | 2 -- > > 1 file changed, 2 deletions(-) > > > > diff --git a/fs/ubifs/gc.c b/fs/ubifs/gc.c > > index 7cc22d7317ea..465beea52176 100644 > > --- a/fs/ubifs/gc.c > > +++ b/fs/ubifs/gc.c > > @@ -899,8 +899,6 @@ int ubifs_gc_start_commit(struct ubifs_info *c) > > err = -ENOMEM; > > goto out; > > } > Hi Colin, > I just found out about it today thanks to your patch. Commit c770cd5190ba > ("ubifs: fix an IS_ERR() vs NULL check") did import a new problem that > ubifs_gc_start_commit() may return -ENOMEM while syncing fs. > I guess ubifs_fast_find_frdi_idx() return NULL pointer is the termination > condition in while-loop, which means we cannot get a freeable index LEB in > ubifs_gc_start_commit(). Ugh... I'm so sorry. My patch was clearly wrong. I've tried before to add a Smatch check which warns about duplicative NULL checks, but I think this gives me a new idea to try. Hopefully, it will prevent this in the future. Yeah, and it's my check which needs to be deleted, not the other one. regards, dan carpenter
----- Ursprüngliche Mail ----- > I just found out about it today thanks to your patch. Commit > c770cd5190ba ("ubifs: fix an IS_ERR() vs NULL check") did import a new > problem that ubifs_gc_start_commit() may return -ENOMEM while syncing fs. > I guess ubifs_fast_find_frdi_idx() return NULL pointer is the > termination condition in while-loop, which means we cannot get a > freeable index LEB in ubifs_gc_start_commit(). Good catch! :-) >> - if (!lp) >> - break; >> idx_gc = kmalloc(sizeof(struct ubifs_gced_idx_leb), GFP_NOFS); >> if (!idx_gc) { >> err = -ENOMEM; >> > BTW, the following modifications may be what you want? > diff --git a/fs/ubifs/gc.c b/fs/ubifs/gc.c > index 7cc22d7317ea..b1f276599b04 100644 > --- a/fs/ubifs/gc.c > +++ b/fs/ubifs/gc.c > @@ -895,10 +895,6 @@ int ubifs_gc_start_commit(struct ubifs_info *c) > /* Record index freeable LEBs for unmapping after commit */ > while (1) { > lp = ubifs_fast_find_frdi_idx(c); > - if (!lp) { > - err = -ENOMEM; > - goto out; > - } > if (!lp) > break; > idx_gc = kmalloc(sizeof(struct ubifs_gced_idx_leb), > GFP_NOFS); I'll drop Dan's patch from -next. Do you want to send a followup patch which removes the in vain check? Thanks, //richard
----- Ursprüngliche Mail ----- > Ugh... I'm so sorry. My patch was clearly wrong. I've tried before to No need to worry. :) > add a Smatch check which warns about duplicative NULL checks, but I > think this gives me a new idea to try. Hopefully, it will prevent this > in the future. Sounds great! Thanks, //richard
diff --git a/fs/ubifs/gc.c b/fs/ubifs/gc.c index 7cc22d7317ea..465beea52176 100644 --- a/fs/ubifs/gc.c +++ b/fs/ubifs/gc.c @@ -899,8 +899,6 @@ int ubifs_gc_start_commit(struct ubifs_info *c) err = -ENOMEM; goto out; } - if (!lp) - break; idx_gc = kmalloc(sizeof(struct ubifs_gced_idx_leb), GFP_NOFS); if (!idx_gc) { err = -ENOMEM;