Re: r188786 introduces a may-be-used-uninitialized warnings in expmed.c

2012-07-05 Thread Richard Guenther
On Wed, Jul 4, 2012 at 5:17 PM, Steven Bosscher wrote: > On Wed, Jul 4, 2012 at 4:25 PM, Uros Bizjak wrote: >> Hello! >> >>> Since r188786, expmed.c has this code: >> >>> This results in warnings for expmed.c during bootstrap on >>> powerpc64-unknown-linux-gnu: >>> >>> ../../trunk/gcc/expmed.c: I

Re: r188786 introduces a may-be-used-uninitialized warnings in expmed.c

2012-07-04 Thread Steven Bosscher
On Wed, Jul 4, 2012 at 4:25 PM, Uros Bizjak wrote: > Hello! > >> Since r188786, expmed.c has this code: > >> This results in warnings for expmed.c during bootstrap on >> powerpc64-unknown-linux-gnu: >> >> ../../trunk/gcc/expmed.c: In function ‘rtx_def* expand_mult(machine_mode, >> rtx, >> rtx, rt

Re: r188786 introduces a may-be-used-uninitialized warnings in expmed.c

2012-07-04 Thread Uros Bizjak
Hello! > Since r188786, expmed.c has this code: > This results in warnings for expmed.c during bootstrap on > powerpc64-unknown-linux-gnu: > > ../../trunk/gcc/expmed.c: In function ‘rtx_def* expand_mult(machine_mode, rtx, > rtx, rtx, int)’: > ../../trunk/gcc/expmed.c:3215:7: warning: ‘is_neg’ may

r188786 introduces a may-be-used-uninitialized warnings in expmed.c

2012-07-03 Thread Steven Bosscher
Hello, Since r188786, expmed.c has this code: bool is_neg; (...) if (CONST_INT_P (scalar_op1)) { coeff = INTVAL (scalar_op1); is_neg = coeff < 0; } else if (CONST_DOUBLE_P (scalar_op1)) { if (CONST_DOUBLE_HIGH (scalar_op1) ==