Mail Archives: pgcc/1999/05/26/15:13:53
This does NOT issue an internal compiler error when using:
gcc-2.8.1
egcs-2.91.66
egcc-2.91.60
--
Jeffrey Hundstad
On 26 May, Ronald de Man wrote:
> Hello,
>
> Compiling the following piece of code with optimization -O or higher,
> leads to an internal compiler error with pgcc-1.1.3:
>
> -----bla.c-----
>
> long long abc;
> long long G(int);
> void bla(int);
>
> struct {
> int h;
> } k;
>
> void R(int t)
> {
> int a,b;
>
> for(a=1;a<=2;a++) {
> if(a)
> k.h^=1;
> bla(4 * (int)(abc >> 60));
> if (G(100) != (abc & 1) )
> b+=G(b)*t;
> }
> }
>
> ---------------
>
> $ gcc -c -O bla.c
> bla.c: In function `R':
> bla.c:20: internal error--insn does not satisfy its constraints:
> (insn:HI 38 111 39 (set (reg:SI 1 %edx)
> (subreg:SI (ashiftrt:DI (reg:DI 0 %eax)
> (const_int 60)) 0)) 253 {ashrdi3_const_int_subreg} (insn_list 34 (nil))
> (expr_list:REG_DEAD (mem:DI (symbol_ref:SI ("abc")))
> (nil)))
> ../../pgcc-1.1.3/gcc/toplev.c:1438: Internal compiler error in function fatal_insn
>
> pgcc-1.1.1 gives a similar error:
>
> $ gcc -Vpgcc-2.91.60 -bi586-pc-linux-gnu -c -O bla.c
> bla.c: In function `R':
> bla.c:20: internal error--insn does not satisfy its constraints:
> (insn:HI 38 106 39 (set (reg:SI 1 %edx)
> (subreg:SI (ashiftrt:DI (reg:DI 0 %eax)
> (const_int 60)) 0)) 242 {ashrdi3_const_int_subreg} (insn_list 34 (nil))
> (expr_list:REG_DEAD (mem:DI (symbol_ref:SI ("abc")))
> (nil)))
> ../../egcs-1.1.1/gcc/toplev.c:1431: Internal compiler error in function fatal_insn
>
>
> Ronald
>
- Raw text -