This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: fix math wrt volatile-bitfields vs C++ model


On Wed, Oct 29, 2014 at 6:24 AM, DJ Delorie <dj@redhat.com> wrote:
>
>> Looks ok to me, but can you add a testcase please?
>>
>> Also check if 4.9 is affected.
>
> Sorry for the delay, this finally made it back to the top of my to-do
> list.  Testcase included which fails without and passes with this
> patch.  4.9 is affected and the same patch fixes it.  Tested on
> rx-elf, x86 32/64, and arm32.

Ok.  For the branch please wait until after 4.9.2 is out.

Thanks,
Richard.

> 2014-10-29  DJ Delorie  <dj@redhat.com>
>
>         * expmed.c (strict_volatile_bitfield_p): Fix off-by-one error.
>
> 2014-10-29  DJ Delorie  <dj@redhat.com>
>
>         * gcc.dg/20141029-1.c: New.
>
>
> Index: expmed.c
> ===================================================================
> --- expmed.c    (revision 216811)
> +++ expmed.c    (working copy)
> @@ -454,13 +454,13 @@ strict_volatile_bitfield_p (rtx op0, uns
>           && bitnum % GET_MODE_ALIGNMENT (fieldmode) + bitsize > modesize))
>      return false;
>
>    /* Check for cases where the C++ memory model applies.  */
>    if (bitregion_end != 0
>        && (bitnum - bitnum % modesize < bitregion_start
> -         || bitnum - bitnum % modesize + modesize > bitregion_end))
> +         || bitnum - bitnum % modesize + modesize - 1 > bitregion_end))
>      return false;
>
>    return true;
>  }
>
>  /* Return true if OP is a memory and if a bitfield of size BITSIZE at
>
> Index: testsuite/gcc.dg/20141029-1.c
> ===================================================================
> --- testsuite/gcc.dg/20141029-1.c       (revision 0)
> +++ testsuite/gcc.dg/20141029-1.c       (revision 0)
> @@ -0,0 +1,28 @@
> +/* { dg-do compile } */
> +/* { dg-options "-fstrict-volatile-bitfields -fdump-rtl-final" } */
> +
> +#define PERIPH (*(volatile struct system_periph *)0x81234)
> +
> +struct system_periph {
> +  union {
> +    unsigned short WORD;
> +    struct {
> +      unsigned short a:1;
> +      unsigned short b:1;
> +      unsigned short  :5;
> +      unsigned short c:1;
> +      unsigned short  :8;
> +    } BIT;
> +  } ALL;
> +};
> +
> +void
> +foo()
> +{
> +  while (1)
> +    {
> +      PERIPH.ALL.BIT.a = 1;
> +    }
> +}
> +/* { dg-final { scan-rtl-dump-times "mem/v(/.)*:HI" 4 "final" } } */
> +/* { dg-final { cleanup-rtl-dump "final" } } */


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]