Message ID | 20170616142144.4244-1-richard@nod.at |
---|---|
State | Accepted |
Delegated to: | Richard Weinberger |
Headers | show |
On Fri, 16 Jun 2017 16:21:44 +0200 Richard Weinberger <richard@nod.at> wrote: > When UBIFS prepares data structures which will be written to the MTD it > ensues that their lengths are multiple of 8. Since it uses kmalloc() the > padded bytes are left uninitialized and we leak a few bytes of kernel > memory to the MTD. > To make sure that all bytes are initialized, let's switch to kzalloc(). > Kzalloc() is fine in this case because the buffers are not huge and in > the IO path the performance bottleneck is anyway the MTD. > > Cc: stable@vger.kernel.org > Fixes: 1e51764a3c2a ("UBIFS: add new flash file system") > Signed-off-by: Richard Weinberger <richard@nod.at> Reviewed-by: Boris Brezillon <boris.brezillon@free-electrons.com> > --- > fs/ubifs/journal.c | 10 +++++----- > 1 file changed, 5 insertions(+), 5 deletions(-) > > diff --git a/fs/ubifs/journal.c b/fs/ubifs/journal.c > index 294519b98874..981a7ea86674 100644 > --- a/fs/ubifs/journal.c > +++ b/fs/ubifs/journal.c > @@ -574,7 +574,7 @@ int ubifs_jnl_update(struct ubifs_info *c, const struct inode *dir, > /* Make sure to also account for extended attributes */ > len += host_ui->data_len; > > - dent = kmalloc(len, GFP_NOFS); > + dent = kzalloc(len, GFP_NOFS); > if (!dent) > return -ENOMEM; > > @@ -967,7 +967,7 @@ int ubifs_jnl_xrename(struct ubifs_info *c, const struct inode *fst_dir, > if (twoparents) > len += plen; > > - dent1 = kmalloc(len, GFP_NOFS); > + dent1 = kzalloc(len, GFP_NOFS); > if (!dent1) > return -ENOMEM; > > @@ -1117,7 +1117,7 @@ int ubifs_jnl_rename(struct ubifs_info *c, const struct inode *old_dir, > len = aligned_dlen1 + aligned_dlen2 + ALIGN(ilen, 8) + ALIGN(plen, 8); > if (move) > len += plen; > - dent = kmalloc(len, GFP_NOFS); > + dent = kzalloc(len, GFP_NOFS); > if (!dent) > return -ENOMEM; > > @@ -1500,7 +1500,7 @@ int ubifs_jnl_delete_xattr(struct ubifs_info *c, const struct inode *host, > hlen = host_ui->data_len + UBIFS_INO_NODE_SZ; > len = aligned_xlen + UBIFS_INO_NODE_SZ + ALIGN(hlen, 8); > > - xent = kmalloc(len, GFP_NOFS); > + xent = kzalloc(len, GFP_NOFS); > if (!xent) > return -ENOMEM; > > @@ -1607,7 +1607,7 @@ int ubifs_jnl_change_xattr(struct ubifs_info *c, const struct inode *inode, > aligned_len1 = ALIGN(len1, 8); > aligned_len = aligned_len1 + ALIGN(len2, 8); > > - ino = kmalloc(aligned_len, GFP_NOFS); > + ino = kzalloc(aligned_len, GFP_NOFS); > if (!ino) > return -ENOMEM; >
diff --git a/fs/ubifs/journal.c b/fs/ubifs/journal.c index 294519b98874..981a7ea86674 100644 --- a/fs/ubifs/journal.c +++ b/fs/ubifs/journal.c @@ -574,7 +574,7 @@ int ubifs_jnl_update(struct ubifs_info *c, const struct inode *dir, /* Make sure to also account for extended attributes */ len += host_ui->data_len; - dent = kmalloc(len, GFP_NOFS); + dent = kzalloc(len, GFP_NOFS); if (!dent) return -ENOMEM; @@ -967,7 +967,7 @@ int ubifs_jnl_xrename(struct ubifs_info *c, const struct inode *fst_dir, if (twoparents) len += plen; - dent1 = kmalloc(len, GFP_NOFS); + dent1 = kzalloc(len, GFP_NOFS); if (!dent1) return -ENOMEM; @@ -1117,7 +1117,7 @@ int ubifs_jnl_rename(struct ubifs_info *c, const struct inode *old_dir, len = aligned_dlen1 + aligned_dlen2 + ALIGN(ilen, 8) + ALIGN(plen, 8); if (move) len += plen; - dent = kmalloc(len, GFP_NOFS); + dent = kzalloc(len, GFP_NOFS); if (!dent) return -ENOMEM; @@ -1500,7 +1500,7 @@ int ubifs_jnl_delete_xattr(struct ubifs_info *c, const struct inode *host, hlen = host_ui->data_len + UBIFS_INO_NODE_SZ; len = aligned_xlen + UBIFS_INO_NODE_SZ + ALIGN(hlen, 8); - xent = kmalloc(len, GFP_NOFS); + xent = kzalloc(len, GFP_NOFS); if (!xent) return -ENOMEM; @@ -1607,7 +1607,7 @@ int ubifs_jnl_change_xattr(struct ubifs_info *c, const struct inode *inode, aligned_len1 = ALIGN(len1, 8); aligned_len = aligned_len1 + ALIGN(len2, 8); - ino = kmalloc(aligned_len, GFP_NOFS); + ino = kzalloc(aligned_len, GFP_NOFS); if (!ino) return -ENOMEM;
When UBIFS prepares data structures which will be written to the MTD it ensues that their lengths are multiple of 8. Since it uses kmalloc() the padded bytes are left uninitialized and we leak a few bytes of kernel memory to the MTD. To make sure that all bytes are initialized, let's switch to kzalloc(). Kzalloc() is fine in this case because the buffers are not huge and in the IO path the performance bottleneck is anyway the MTD. Cc: stable@vger.kernel.org Fixes: 1e51764a3c2a ("UBIFS: add new flash file system") Signed-off-by: Richard Weinberger <richard@nod.at> --- fs/ubifs/journal.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-)