drivers/staging/dgrp/dgrp_tty.c:3177 dgrp_tty_init() error: potential null deref

Hi Bill,

FYI, there are new smatch warnings show up in

tree:   git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty.git tty-next
head:   0c57dfcc6c1d037243c2f8fbf62eab3633326ec0
commit: 7b6d45c211a401a9bdeebfa96f8a4c811bd3eeaf [9/39] staging: dgrp: add dgrp to the build

It seems that the dgrp_dpa_ops.c ones are false warnings, while the 'potential null deference' ones are worth looking at.

+ drivers/staging/dgrp/dgrp_dpa_ops.c:377 dgrp_dpa_ioctl() warn: 'vpd' puts 516 bytes on stack
--
+ drivers/staging/dgrp/dgrp_net_ops.c:277 dgrp_input() error: we previously assumed 'ld' could be null (see line 238)
+ drivers/staging/dgrp/dgrp_net_ops.c:3159 dgrp_receive() error: strncpy() 'nd->nd_ps_desc' too small (101 vs 988)
+ drivers/staging/dgrp/dgrp_net_ops.c:3160 dgrp_receive() error: buffer overflow 'nd->nd_ps_desc' 101 <= 988
--
+ drivers/staging/dgrp/dgrp_specproc.c:273 register_proc_table() error: we previously assumed 'table->child' could be null (see line 263) --
+ drivers/staging/dgrp/dgrp_tty.c:2341 dgrp_set_custom_speed() info: ignoring unreachable code.
+ drivers/staging/dgrp/dgrp_tty.c:3177 dgrp_tty_init() error: potential null dereference 'nd->nd_serial_ttdriver'. (alloc_tty_driver returns null)
+ drivers/staging/dgrp/dgrp_tty.c:3236 dgrp_tty_init() error: potential null dereference 'nd->nd_callout_ttdriver'. (alloc_tty_driver returns null)
+ drivers/staging/dgrp/dgrp_tty.c:3273 dgrp_tty_init() error: potential null dereference 'nd->nd_xprint_ttdriver'. (alloc_tty_driver returns null)

---
0-DAY kernel build testing backend Open Source Technology Center
Fengguang Wu, Yuanhan Liu Intel Corporation

This one looks fine, if ld is NULL then len is set to 0. ld is then only dereferenced if len != 0

It seems weird where smatch is seeing the NULL test vs. the call to register_proc_table(), but it is correct that register_proc_table will do bad things if the first arg is a NULL. I'll submit a patch for it.

Yes, the return value of alloc_tty_driver needs to be checked. I'll fix these.

Bill

I think thats the tip of the iceberg at best 8)

Alan