Not only memcpy() is faster than pointer copy, but also let code more
clearer and simple, which can avoid compiling warning (the original
implementation copy registers by exceeding member array border).
The related warning (with allmodconfig under tile):
CC arch/tile/kernel/kgdb.o
arch/tile/kernel/kgdb.c: In function 'sleeping_thread_to_gdb_regs':
arch/tile/kernel/kgdb.c:140:31: warning: iteration 53u invokes undefined behavior [-Waggressive-loop-optimizations]
*(ptr++) = thread_regs->regs[reg];
^
arch/tile/kernel/kgdb.c:139:2: note: containing loop
for (reg = 0; reg <= TREG_LAST_GPR; reg++)
^
Signed-off-by: Chen Gang <gang.chen.5i5j@gmail.com>
Signed-off-by: Chris Metcalf <cmetcalf@ezchip.com>
void
sleeping_thread_to_gdb_regs(unsigned long *gdb_regs, struct task_struct *task)
{
- int reg;
struct pt_regs *thread_regs;
- unsigned long *ptr = gdb_regs;
if (task == NULL)
return;
memset(gdb_regs, 0, NUMREGBYTES);
thread_regs = task_pt_regs(task);
- for (reg = 0; reg <= TREG_LAST_GPR; reg++)
- *(ptr++) = thread_regs->regs[reg];
-
+ memcpy(gdb_regs, thread_regs, TREG_LAST_GPR * sizeof(unsigned long));
gdb_regs[TILEGX_PC_REGNUM] = thread_regs->pc;
gdb_regs[TILEGX_FAULTNUM_REGNUM] = thread_regs->faultnum;
}