RISC-V: Avoid dereferening NULL regs in die()
authorPalmer Dabbelt <palmer@rivosinc.com>
Tue, 20 Sep 2022 20:00:37 +0000 (13:00 -0700)
committerPalmer Dabbelt <palmer@rivosinc.com>
Thu, 13 Oct 2022 05:02:49 +0000 (22:02 -0700)
I don't think we can actually die() without a regs pointer, but the
compiler was warning about a NULL check after a dereference.  It seems
prudent to just avoid the possibly-NULL dereference, given that when
die()ing the system is already toast so who knows how we got there.

Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://lore.kernel.org/r/20220920200037.6727-1-palmer@rivosinc.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
arch/riscv/kernel/traps.c

index 635e6ec..f3e96d6 100644 (file)
@@ -33,6 +33,7 @@ void die(struct pt_regs *regs, const char *str)
 {
        static int die_counter;
        int ret;
+       long cause;
 
        oops_enter();
 
@@ -42,11 +43,13 @@ void die(struct pt_regs *regs, const char *str)
 
        pr_emerg("%s [#%d]\n", str, ++die_counter);
        print_modules();
-       show_regs(regs);
+       if (regs)
+               show_regs(regs);
 
-       ret = notify_die(DIE_OOPS, str, regs, 0, regs->cause, SIGSEGV);
+       cause = regs ? regs->cause : -1;
+       ret = notify_die(DIE_OOPS, str, regs, 0, cause, SIGSEGV);
 
-       if (regs && kexec_should_crash(current))
+       if (kexec_should_crash(current))
                crash_kexec(regs);
 
        bust_spinlocks(0);