Re: [Qemu-block] [PATCH 4/4] Remove unnecessary variables for function return value

[Eric Blake]

On 03/22/2018 11:12 AM, Laurent Vivier wrote:
> Re-run Coccinelle script scripts/coccinelle/return_directly.cocci
>
> Signed-off-by: Laurent Vivier <address@hidden>
> ---

> +++ b/hw/arm/exynos4210.c
> @@ -156,12 +156,7 @@ void exynos4210_write_secondary(ARMCPU *cpu,
>   
>   static uint64_t exynos4210_calc_affinity(int cpu)
>   {
> -    uint64_t mp_affinity;
> -
> -    /* Exynos4210 has 0x9 as cluster ID */
> -    mp_affinity = (0x9 << ARM_AFF1_SHIFT) | cpu;
> -
> -    return mp_affinity;
> +    return (0x9 << ARM_AFF1_SHIFT) | cpu;

Unchanged by this rewrite, but since this is converting a signed 32-bit 
int to an unsigned 64-bit value, are we sure that the upper 32 bits are 
always set correctly?  (Using unsigned values earlier in the expression 
would require less head-scratching on whether it is correct).  Any 
changes should be a separate fix by the file's maintainer.


> +++ b/hw/misc/mos6522.c
> @@ -176,12 +176,8 @@ static void mos6522_set_sr_int(MOS6522State *s)
>   
>   static uint64_t mos6522_get_counter_value(MOS6522State *s, MOS6522Timer *ti)
>   {
> -    uint64_t d;
> -
> -    d = muldiv64(qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) - ti->load_time,
> +    return muldiv64(qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) - ti->load_time,
>                    ti->frequency, NANOSECONDS_PER_SECOND);

Coccinelle missed that indentation is now off here.

> +++ b/hw/ppc/pnv_lpc.c
> @@ -125,25 +125,15 @@ static int pnv_lpc_dt_xscom(PnvXScomInterface *dev, void 
> *fdt, int xscom_offset)
>   static bool opb_read(PnvLpcController *lpc, uint32_t addr, uint8_t *data,
>                        int sz)
>   {
> -    bool success;
> -
> -    /* XXX Handle access size limits and FW read caching here */
> -    success = !address_space_rw(&lpc->opb_as, addr, MEMTXATTRS_UNSPECIFIED,
> +    return !address_space_rw(&lpc->opb_as, addr, MEMTXATTRS_UNSPECIFIED,
>                                   data, sz, false);

and here.

> +++ b/target/xtensa/translate.c
> @@ -1272,11 +1272,8 @@ XtensaOpcodeOps *
>   xtensa_find_opcode_ops(const XtensaOpcodeTranslators *t,
>                          const char *name)
>   {
> -    XtensaOpcodeOps *ops;
> -
> -    ops = bsearch(name, t->opcode, t->num_opcodes,
> +    return bsearch(name, t->opcode, t->num_opcodes,
>                     sizeof(XtensaOpcodeOps), compare_opcode_ops);

and here


--
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3266
Virtualization:  qemu.org | libvirt.org