Re: [PATCH] powerpc/sstep: Fix array out of bound warning

["Naveen N. Rao" <naveen.n.rao@linux.vnet.ibm.com>]

On 2021/01/15 11:46AM, Ravi Bangoria wrote:
> Compiling kernel with -Warray-bounds throws below warning:
> 
>   In function 'emulate_vsx_store':
>   warning: array subscript is above array bounds [-Warray-bounds]
>   buf.d[2] = byterev_8(reg->d[1]);
>   ~~~~~^~~
>   buf.d[3] = byterev_8(reg->d[0]);
>   ~~~~~^~~
> 
> Fix it by converting local variable 'union vsx_reg buf' into an array.
> Also consider function argument 'union vsx_reg *reg' as array instead
> of pointer because callers are actually passing an array to it.

I think you should change the function prototype to reflect this.

However, while I agree with this change in principle, it looks to be a 
lot of code churn for a fairly narrow use. Perhaps we should just 
address the specific bug. Something like the below (not tested)?

@@ -818,13 +818,15 @@ void emulate_vsx_store(struct instruction_op *op, const union vsx_reg *reg,
                        break;
                if (rev) {
                        /* reverse 32 bytes */
-                       buf.d[0] = byterev_8(reg->d[3]);
-                       buf.d[1] = byterev_8(reg->d[2]);
-                       buf.d[2] = byterev_8(reg->d[1]);
-                       buf.d[3] = byterev_8(reg->d[0]);
-                       reg = &buf;
+                       union vsx_reg buf32[2];
+                       buf32[0].d[0] = byterev_8(reg[1].d[1]);
+                       buf32[0].d[1] = byterev_8(reg[1].d[0]);
+                       buf32[1].d[0] = byterev_8(reg[0].d[1]);
+                       buf32[1].d[1] = byterev_8(reg[0].d[0]);
+                       memcpy(mem, buf32, size);
+               } else {
+                       memcpy(mem, reg, size);
                }
-               memcpy(mem, reg, size);
                break;
        case 16:
                /* stxv, stxvx, stxvl, stxvll */


- Naveen