[bug report] [MIPS] MT: Improved multithreading support.

[Dan Carpenter <dan.carpenter@oracle.com>]

[ This is ancient code, but the bug is straight forward -- dan ]

Hello Ralf Baechle,

The patch 41c594ab65fc: "[MIPS] MT: Improved multithreading support."
from Apr 5, 2006, leads to the following static checker warning:

	arch/mips/kernel/smp-mt.c:168 vsmp_boot_secondary()
	warn: potential pointer math issue

arch/mips/kernel/smp-mt.c
   142  static int vsmp_boot_secondary(int cpu, struct task_struct *idle)
   143  {
   144          struct thread_info *gp = task_thread_info(idle);
                ^^^^^^^^^^^^^^^^^^^^^^
gp is a thread_info struct pointer.

   145          dvpe();
   146          set_c0_mvpcontrol(MVPCONTROL_VPC);
   147  
   148          settc(cpu);
   149  
   150          /* restart */
   151          write_tc_c0_tcrestart((unsigned long)&smp_bootstrap);
   152  
   153          /* enable the tc this vpe/cpu will be running */
   154          write_tc_c0_tcstatus((read_tc_c0_tcstatus() & ~TCSTATUS_IXMT) | TCSTATUS_A);
   155  
   156          write_tc_c0_tchalt(0);
   157  
   158          /* enable the VPE */
   159          write_vpe_c0_vpeconf0(read_vpe_c0_vpeconf0() | VPECONF0_VPA);
   160  
   161          /* stack pointer */
   162          write_tc_gpr_sp( __KSTK_TOS(idle));
   163  
   164          /* global pointer */
   165          write_tc_gpr_gp((unsigned long)gp);
   166  
   167          flush_icache_range((unsigned long)gp,
   168                             (unsigned long)(gp + sizeof(struct thread_info)));
                                                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
This should almost certainly be "gp + 1" because of pointer math.

   169  
   170          /* finally out of configuration and into chaos */
   171          clear_c0_mvpcontrol(MVPCONTROL_VPC);
   172  
   173          evpe(EVPE_ENABLE);
   174  
   175          return 0;
   176  }

regards,
dan carpenter