Re: [Qemu-devel] [PATCH v9 05/15] target-or32: Add int instruction helpers

[Jia Liu]

Hi Max

On Fri, Jul 13, 2012 at 5:18 PM, Max Filippov <address@hidden> wrote:
> Hi.
>
> On Fri, Jul 13, 2012 at 8:34 AM, Jia Liu <address@hidden> wrote:
>> Add OpenRISC int instruction helpers.
>>
>> Signed-off-by: Jia Liu <address@hidden>
>> ---
>>  target-openrisc/Makefile.objs |    2 +-
>>  target-openrisc/helper.h      |    5 +++
>>  target-openrisc/int_helper.c  |   87 
>> +++++++++++++++++++++++++++++++++++++++++
>>  3 files changed, 93 insertions(+), 1 deletion(-)
>>  create mode 100644 target-openrisc/int_helper.c
>>
>> diff --git a/target-openrisc/Makefile.objs b/target-openrisc/Makefile.objs
>> index 52d0158..e2a3715 100644
>> --- a/target-openrisc/Makefile.objs
>> +++ b/target-openrisc/Makefile.objs
>> @@ -1,3 +1,3 @@
>>  obj-$(CONFIG_SOFTMMU) += machine.o
>>  obj-y += cpu.o exception.o interrupt.o mmu.o translate.o
>> -obj-y += exception_helper.o interrupt_helper.o mmu_helper.o
>> +obj-y += exception_helper.o int_helper.o interrupt_helper.o mmu_helper.o
>> diff --git a/target-openrisc/helper.h b/target-openrisc/helper.h
>> index 4e2a49f..c772951 100644
>> --- a/target-openrisc/helper.h
>> +++ b/target-openrisc/helper.h
>> @@ -22,6 +22,11 @@
>>  /* exception */
>>  DEF_HELPER_FLAGS_2(exception, 0, void, env, i32)
>>
>> +/* int */
>> +DEF_HELPER_FLAGS_1(ff1, 0, tl, tl)
>> +DEF_HELPER_FLAGS_1(fl1, 0, tl, tl)
>> +DEF_HELPER_FLAGS_3(mul32, 0, tl, env, tl, tl)
>> +
>>  /* interrupt */
>>  DEF_HELPER_FLAGS_1(rfe, 0, void, env)
>>
>> diff --git a/target-openrisc/int_helper.c b/target-openrisc/int_helper.c
>> new file mode 100644
>> index 0000000..45a9bfd
>> --- /dev/null
>> +++ b/target-openrisc/int_helper.c
>> @@ -0,0 +1,87 @@
>> +/*
>> + * OpenRISC int helper routines
>> + *
>> + * Copyright (c) 2011-2012 Jia Liu <address@hidden>
>> + *                         Feng Gao <address@hidden>
>> + *
>> + * This library is free software; you can redistribute it and/or
>> + * modify it under the terms of the GNU Lesser General Public
>> + * License as published by the Free Software Foundation; either
>> + * version 2 of the License, or (at your option) any later version.
>> + *
>> + * This library is distributed in the hope that it will be useful,
>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
>> + * Lesser General Public License for more details.
>> + *
>> + * You should have received a copy of the GNU Lesser General Public
>> + * License along with this library; if not, see 
>> <http://www.gnu.org/licenses/>.
>> + */
>> +
>> +#include "cpu.h"
>> +#include "helper.h"
>> +#include "exception.h"
>> +
>> +target_ulong HELPER(ff1)(target_ulong x)
>> +{
>> +    target_ulong n = 0;
>> +
>> +    if (x == 0) {
>> +        return 0;
>> +    }
>> +
>> +    for (n = 32; x; n--) {
>> +        x <<= 1;
>> +    }
>> +    return n+1;
>> +}
>
> This may be implemented more efficiently with "return x ? ctz32(x) + 1 : 
> 0;"...
>
>> +target_ulong HELPER(fl1)(target_ulong x)
>> +{
>> +    target_ulong n = 0;
>> +
>> +    if (x == 0) {
>> +        return 0;
>> +    }
>> +
>> +    for (n = 0; x; n++) {
>> +        x >>= 1;
>> +    }
>> +    return n;
>> +}
>
> ...and this with "return 32 - clz32(x);"
>

Thank you very much! And I made them like this:
target_ulong HELPER(ff1)(target_ulong x)
{
/*#ifdef TARGET_OPENRISC64
    return x ? ctz64(x) + 1 : 0;
#else*/
    return x ? ctz32(x) + 1 : 0;
/*#endif*/
}

target_ulong HELPER(fl1)(target_ulong x)
{
/* not used yet, open it when we need or64.  */
/*#ifdef TARGET_OPENRISC64
    return 64 - clz64(x);
#else*/
    return 32 - clz32(x);
/*#endif*/
}

I have to comment out all of the or64 code,for now.
Is it OK?

>> +target_ulong HELPER(mul32)(CPUOpenRISCState *env,
>> +                           target_ulong ra, target_ulong rb)
>> +{
>> +    uint64_t result;
>> +    target_ulong high, cy;
>> +
>> +    OpenRISCCPU *cpu = OPENRISC_CPU(ENV_GET_CPU(env));
>> +
>> +    result = ra * rb;
>
> I believe that this multiplication should have been written as
>
>         result = (uint64_t)ra * rb;
>
> in order to get 64 x 64 => 64 instead of 32 x 32 => 32 => 64.
>

Thank you!
Is this better?
uint32_t HELPER(mul32)(CPUOpenRISCState *env,
                           uint32_t ra, uint32_t rb)
{
    uint64_t result;
    uint32_t high, cy;

    OpenRISCCPU *cpu = OPENRISC_CPU(ENV_GET_CPU(env));

    result = (uint64_t)ra * rb;

I made it a pure 32bit mul.

>> +    /* regisiers in or32 is 32bit, so 32 is NOT a magic number.
>> +       or64 is not handled in this function, and not implement yet,
>> +       TARGET_LONG_BITS for or64 is 64, it will break this function,
>> +       so, we didn't use TARGET_LONG_BITS here.  */
>> +    high = result >> 32;
>> +    cy = result >> (32 - 1);
>> +
>> +    if ((cy & 0x1) == 0x0) {
>> +        if (high == 0x0) {
>> +            return result;
>> +        }
>> +    }
>> +
>> +    if ((cy & 0x1) == 0x1) {
>> +        if (high == 0xffffffff) {
>> +            return result;
>> +        }
>> +    }
>> +
>> +    cpu->env.sr |= (SR_OV | SR_CY);
>> +    if (cpu->env.sr & SR_OVE) {
>> +        raise_exception(cpu, EXCP_RANGE);
>> +    }
>> +
>> +    return result;
>> +}
>> --
>> 1.7.9.5
>
> --
> Thanks.
> -- Max

Regards,
Jia