On Tue, Mar 28, 2017 at 2:40 PM, Dibyendu Majumdar
<mobile@xxxxxxxxxxxxxxx> wrote:
> I am getting a failure in following test when simplifications are enabled:
>
> extern int printf(const char *, ...);
>
> int main(void)
> {
> struct{
> int twobit:2;
> int :1;
> int threebit:3;
> unsigned int onebit:1;
> } s3;
>
> s3.onebit = 1;
> if(s3.onebit != 1){
> printf("Be especially careful with 1-bit fields! %d\n", (int) s3.onebit);
> return 1;
> }
> return 0;
> }
>
> The output (truncated) from linearizer without simplications is this:
>
> main:
> .L0:
> <entry-point>
> load.32 %r1 <- 0[s3]
> shl.32 %r2 <- $1, $6
> and.32 %r3 <- %r1, $-65
> or.32 %r4 <- %r3, %r2
> store.32 %r4 -> 0[s3]
> load.32 %r5 <- 0[s3]
> lsr.32 %r6 <- %r5, $6
> cast.32 %r7 <- (1) %r6
> setne.32 %r8 <- %r7, $1
> br %r8, .L1, .L2
>
> But if simplifications are on then we get:
>
> main:
> .L0:
> <entry-point>
> and.32 %r3 <- %r4, $-65
> or.32 %r4 <- %r3, $64
> lsr.32 %r6 <- %r4, $6
> cast.32 %r7 <- (1) %r6
> setne.32 %r8 <- %r7, $1
> br %r8, .L1, .L3
Are you worried about the load of s3 (wich is not initialized)?
Or the store of s3 (wich is local and never used but for .onebit)?
Or the size of %r6 (which is created as 32bit but used as 1bit)?
If the later, you're missing the patch
https://git.kernel.org/pub/scm/devel/sparse/sparse.git/commit/?h=sparse-next&id=522773d089700cce5551860aea3cb93f40b5f3a4
-- Luc
--
To unsubscribe from this list: send the line "unsubscribe linux-sparse" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
