On Tue, Dec 3, 2013 at 1:48 PM, Bernd Edlinger <bernd.edlin...@hotmail.de> wrote: > Hi Jeff, > > please find attached the patch (incl. test cases) for the unaligned read BUG > that I found while investigating > on PR#57748: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57748 > > one test case is this one: > > pr57748-3.c: > /* PR middle-end/57748 */ > /* { dg-do run } */ > /* wrong code in expand_expr_real_1. */ > > #include <stdlib.h> > > extern void abort (void); > > typedef long long V > __attribute__ ((vector_size (2 * sizeof (long long)), may_alias)); > > typedef struct S { V a; V b[0]; } P __attribute__((aligned (1))); > > struct __attribute__((packed)) T { char c; P s; }; > > void __attribute__((noinline, noclone)) > check (P *p) > { > if (p->b[0][0] != 3 || p->b[0][1] != 4) > abort (); > } > > void __attribute__((noinline, noclone)) > foo (struct T *t) > { > V a = { 3, 4 }; > t->s.b[0] = a; > } > > int > main () > { > struct T *t = (struct T *) calloc (128, 1); > > foo (t); > check (&t->s); > > free (t); > return 0; > } > > > and the other one is > pr57748-4.c: > /* PR middle-end/57748 */ > /* { dg-do run } */ > /* wrong code in expand_expr_real_1. */ > > #include <stdlib.h> > > extern void abort (void); > > typedef long long V > __attribute__ ((vector_size (2 * sizeof (long long)), may_alias)); > > typedef struct S { V b[1]; } P __attribute__((aligned (1))); > > struct __attribute__((packed)) T { char c; P s; }; > > void __attribute__((noinline, noclone)) > check (P *p) > { > if (p->b[1][0] != 3 || p->b[1][1] != 4) > abort (); > } > > void __attribute__((noinline, noclone)) > foo (struct T *t) > { > V a = { 3, 4 }; > t->s.b[1] = a; > } > > int > main () > { > struct T *t = (struct T *) calloc (128, 1); > > foo (t); > check (&t->s); > > free (t); > return 0; > } > > > The patch does add a boolean "expand_reference" parameter to expand_expr_real > and > expand_expr_real_1. I pass true when I intend to use the returned memory > context > as an array reference, instead of a value. At places where mis-aligned values > are extracted, > I do not return a register with the extracted mis-aligned value if > expand_reference is true. > When I have a VIEW_CONVERT_EXPR I pay attention to pass down the outer > "expand_reference" > to the inner expand_expr_real call. Expand_reference, is pretty much similar > to the > expand_modifier "EXPAND_MEMORY". > > Boot-strapped and regression-tested on X86_64-pc-linux-gnu (many times). > > Ok for trunk?
It still feels like papering over the underlying issue. Let me have a second (or third?) look. Richard. > > Thanks > Bernd.
