Re: [PATCH bpf-next] bpf: improve the general precision of tnum_mul

[Nandakumar Edamana <nandakumar@xxxxxxxxxxxxxxxx>]

The following change seems to simplify the code and improve the precision:

-                       /* In case LSB(a) is 1 */
-                       u64 itermask = b.value | b.mask;
-                       struct tnum iterprod = TNUM(b.value & ~itermask, 
itermask);
-                       struct tnum acc_1 = tnum_add(acc, iterprod);
-
-                       acc = tnum_union(acc, acc_1);
+                       acc = tnum_union(acc, tnum_add(acc, b)); /* 
tnum_union(acc_0, acc_1) */

I'll check if it can be improved further.

--
Nandakumar Edamana