| Message ID | 20190107171515.4537-1-colin.king@canonical.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | cifs: fix memory leak of an allocated cifs_ntsd structure | expand |
On Mon, Jan 07, 2019 at 05:15:15PM +0000, Colin King wrote: > From: Colin Ian King <colin.king@canonical.com> > > The call to SMB2_queary_acl can allocate memory to pntsd and also > return a failure via a call to SMB2_query_acl (and then query_info). > This occurs when query_info allocates the structure and then in > query_info the call to smb2_validate_and_copy_iov fails. Currently the > failure just returns without kfree'ing pntsd hence causing a memory > leak. Fix this by kfree'ing pntsd before returning. > > Detected by CoverityScan, CID#1457059 ("Resource Leak") > > Fixes: 2f1afe25997f ("cifs: Use smb 2 - 3 and cifsacl mount options getacl functions") > Signed-off-by: Colin Ian King <colin.king@canonical.com> > --- > fs/cifs/smb2ops.c | 4 +++- > 1 file changed, 3 insertions(+), 1 deletion(-) > > diff --git a/fs/cifs/smb2ops.c b/fs/cifs/smb2ops.c > index cf7eb891804f..6d71958ad2cb 100644 > --- a/fs/cifs/smb2ops.c > +++ b/fs/cifs/smb2ops.c > @@ -2238,8 +2238,10 @@ get_smb2_acl_by_fid(struct cifs_sb_info *cifs_sb, > cifs_put_tlink(tlink); > > cifs_dbg(FYI, "%s: rc = %d ACL len %d\n", __func__, rc, *pacllen); > - if (rc) > + if (rc) { > + kfree(pntsd); > return ERR_PTR(rc); > + } This is a layering violation. The memory was allocated in query_info() so it should be freed there instead. Also if the kmalloc() fails in query_info() then it should return -ENOMEM instead of success. This only affects code which calls SMB2_query_acl(). There are two callers. You have fixed one but the other is also buggy because we're returning uninitialized memory in get_smb2_acl_by_path(). regards, dan carpenter
On 08/01/2019 09:42, Dan Carpenter wrote: > On Mon, Jan 07, 2019 at 05:15:15PM +0000, Colin King wrote: >> From: Colin Ian King <colin.king@canonical.com> >> >> The call to SMB2_queary_acl can allocate memory to pntsd and also >> return a failure via a call to SMB2_query_acl (and then query_info). >> This occurs when query_info allocates the structure and then in >> query_info the call to smb2_validate_and_copy_iov fails. Currently the >> failure just returns without kfree'ing pntsd hence causing a memory >> leak. Fix this by kfree'ing pntsd before returning. >> >> Detected by CoverityScan, CID#1457059 ("Resource Leak") >> >> Fixes: 2f1afe25997f ("cifs: Use smb 2 - 3 and cifsacl mount options getacl functions") >> Signed-off-by: Colin Ian King <colin.king@canonical.com> >> --- >> fs/cifs/smb2ops.c | 4 +++- >> 1 file changed, 3 insertions(+), 1 deletion(-) >> >> diff --git a/fs/cifs/smb2ops.c b/fs/cifs/smb2ops.c >> index cf7eb891804f..6d71958ad2cb 100644 >> --- a/fs/cifs/smb2ops.c >> +++ b/fs/cifs/smb2ops.c >> @@ -2238,8 +2238,10 @@ get_smb2_acl_by_fid(struct cifs_sb_info *cifs_sb, >> cifs_put_tlink(tlink); >> >> cifs_dbg(FYI, "%s: rc = %d ACL len %d\n", __func__, rc, *pacllen); >> - if (rc) >> + if (rc) { >> + kfree(pntsd); >> return ERR_PTR(rc); >> + } > > This is a layering violation. The memory was allocated in query_info() > so it should be freed there instead. Also if the kmalloc() fails in > query_info() then it should return -ENOMEM instead of success. > > This only affects code which calls SMB2_query_acl(). There are two > callers. You have fixed one but the other is also buggy because we're > returning uninitialized memory in get_smb2_acl_by_path(). > > regards, > dan carpenter > Eep. very true. NACK to my fixes and I'll sort out a correct fix later. Colin
Yes - good point. Will back this out temporarily, and also update Aurelien's (unrelated patch) to the later version On Tue, Jan 8, 2019 at 3:52 AM Colin Ian King <colin.king@canonical.com> wrote: > > On 08/01/2019 09:42, Dan Carpenter wrote: > > On Mon, Jan 07, 2019 at 05:15:15PM +0000, Colin King wrote: > >> From: Colin Ian King <colin.king@canonical.com> > >> > >> The call to SMB2_queary_acl can allocate memory to pntsd and also > >> return a failure via a call to SMB2_query_acl (and then query_info). > >> This occurs when query_info allocates the structure and then in > >> query_info the call to smb2_validate_and_copy_iov fails. Currently the > >> failure just returns without kfree'ing pntsd hence causing a memory > >> leak. Fix this by kfree'ing pntsd before returning. > >> > >> Detected by CoverityScan, CID#1457059 ("Resource Leak") > >> > >> Fixes: 2f1afe25997f ("cifs: Use smb 2 - 3 and cifsacl mount options getacl functions") > >> Signed-off-by: Colin Ian King <colin.king@canonical.com> > >> --- > >> fs/cifs/smb2ops.c | 4 +++- > >> 1 file changed, 3 insertions(+), 1 deletion(-) > >> > >> diff --git a/fs/cifs/smb2ops.c b/fs/cifs/smb2ops.c > >> index cf7eb891804f..6d71958ad2cb 100644 > >> --- a/fs/cifs/smb2ops.c > >> +++ b/fs/cifs/smb2ops.c > >> @@ -2238,8 +2238,10 @@ get_smb2_acl_by_fid(struct cifs_sb_info *cifs_sb, > >> cifs_put_tlink(tlink); > >> > >> cifs_dbg(FYI, "%s: rc = %d ACL len %d\n", __func__, rc, *pacllen); > >> - if (rc) > >> + if (rc) { > >> + kfree(pntsd); > >> return ERR_PTR(rc); > >> + } > > > > This is a layering violation. The memory was allocated in query_info() > > so it should be freed there instead. Also if the kmalloc() fails in > > query_info() then it should return -ENOMEM instead of success. > > > > This only affects code which calls SMB2_query_acl(). There are two > > callers. You have fixed one but the other is also buggy because we're > > returning uninitialized memory in get_smb2_acl_by_path(). > > > > regards, > > dan carpenter > > > > Eep. very true. NACK to my fixes and I'll sort out a correct fix later. > > Colin
diff --git a/fs/cifs/smb2ops.c b/fs/cifs/smb2ops.c index cf7eb891804f..6d71958ad2cb 100644 --- a/fs/cifs/smb2ops.c +++ b/fs/cifs/smb2ops.c @@ -2238,8 +2238,10 @@ get_smb2_acl_by_fid(struct cifs_sb_info *cifs_sb, cifs_put_tlink(tlink); cifs_dbg(FYI, "%s: rc = %d ACL len %d\n", __func__, rc, *pacllen); - if (rc) + if (rc) { + kfree(pntsd); return ERR_PTR(rc); + } return pntsd; }