Re: [PATCH] misc: mei: client.c: fix some error code problem in mei_cl_write

[Su Hui]

On 2023/11/20 23:03, Dan Carpenter wrote:
> On Mon, Nov 20, 2023 at 04:53:45PM +0800, Su Hui wrote:
> > Clang static analyzer complains that value stored to 'rets' is never
> > read. Remove some useless code, and let 'buf_len = -EOVERFLOW' to make
> > sure we can return '-EOVERFLOW'.
> >
> > mei_msg_hdr_init() return negative error code, rets should be
> > 'PTR_ERR(mei_hdr)' rather than '-PTR_ERR(mei_hdr)'.
> >
> > Fixes: 0cd7c01a60f8 ("mei: add support for mei extended header.")
> > Fixes: 8c8d964ce90f ("mei: move hbuf_depth from the mei device to the hw modules")
> > Signed-off-by: Su Hui <suhui@xxxxxxxxxxxx>
> > ---
> >   drivers/misc/mei/client.c | 6 ++----
> >   1 file changed, 2 insertions(+), 4 deletions(-)
> >
> > diff --git a/drivers/misc/mei/client.c b/drivers/misc/mei/client.c
> > index 9c8fc87938a7..00dac0a47da0 100644
> > --- a/drivers/misc/mei/client.c
> > +++ b/drivers/misc/mei/client.c
> > @@ -2011,7 +2011,7 @@ ssize_t mei_cl_write(struct mei_cl *cl, struct mei_cl_cb *cb, unsigned long time
> >   
> >   	mei_hdr = mei_msg_hdr_init(cb);
> >   	if (IS_ERR(mei_hdr)) {
> > -		rets = -PTR_ERR(mei_hdr);
> > +		rets = PTR_ERR(mei_hdr);
> KTODO: write a static checker rule which complains -PTR_ERR()
>
> This might be complicated because there are parts of networking where
> we store error codes as positive values.  But there is enough context in
> this function to create some sort of warning about this code.  "Mixing
> positive and negative error codes" perhaps?

Maybe add a check in checkpatch.pl is a good idea?

> $ git grep -n '\-PTR_ERR'
> block/partitions/core.c:574:                   disk->disk_name, p, -PTR_ERR(part));
> drivers/infiniband/ulp/ipoib/ipoib_multicast.c:291:                        -PTR_ERR(ah));
> drivers/misc/mei/client.c:2014:         rets = -PTR_ERR(mei_hdr);
> drivers/net/ethernet/intel/igb/igb_main.c:8963:                 unsigned int xdp_res = -PTR_ERR(skb);
> drivers/net/ethernet/intel/igc/igc_main.c:2639:                 unsigned int xdp_res = -PTR_ERR(skb);
> drivers/net/ethernet/intel/ixgbe/ixgbe_main.c:2348:                     unsigned int xdp_res = -PTR_ERR(skb);
> drivers/net/ethernet/stmicro/stmmac/stmmac_main.c:5431:                         unsigned int xdp_res = -PTR_ERR(skb);
> drivers/phy/sunplus/phy-sunplus-usb2.c:278:             ret = -PTR_ERR(phy);
> drivers/scsi/libfc/fc_elsct.c:86:               switch (-PTR_ERR(fp)) {
> drivers/scsi/libfc/fc_lport.c:1081:                  IS_ERR(fp) ? -PTR_ERR(fp) : 0, fc_lport_state(lport),
> fs/ext4/indirect.c:1042:                                ext4_error_inode_block(inode, nr, -PTR_ERR(bh),
> fs/jffs2/background.c:48:                       -PTR_ERR(tsk));
> fs/ntfs/dir.c:92:                               -PTR_ERR(m));
> fs/ntfs/dir.c:312:                              -PTR_ERR(page));
> fs/ntfs/dir.c:643:                              -PTR_ERR(m));
> fs/ntfs/dir.c:790:                              -PTR_ERR(page));
> fs/ntfs/index.c:141:                            -PTR_ERR(m));
> fs/ntfs/index.c:268:                            -PTR_ERR(page));
> fs/ntfs/mft.c:162:      ntfs_error(ni->vol->sb, "Failed with error code %lu.", -PTR_ERR(m));
> fs/ntfs/mft.c:289:                              "mft record, error code %ld.", -PTR_ERR(m));
> net/ipv6/af_inet6.c:852:                        WRITE_ONCE(sk->sk_err_soft, -PTR_ERR(dst));
> net/ipv6/inet6_connection_sock.c:123:           WRITE_ONCE(sk->sk_err_soft, -PTR_ERR(dst));
> tools/lib/bpf/libbpf.c:10044:           errno = -PTR_ERR(ptr);
> tools/lib/bpf/libbpf_internal.h:520:            errno = -PTR_ERR(ret);
> tools/testing/selftests/bpf/prog_tests/btf_dump.c:59:           err = -PTR_ERR(btf);
> tools/testing/selftests/bpf/prog_tests/sk_lookup.c:483:         errno = -PTR_ERR(link);
>
> Quite a few of those were in printks and it might be an opportunity to
> use %pe to print the ENOMEM etc strings instead of the number.

I will try sending some patch for this.

Thanks for your suggestions.

Su Hui