Re: [PATCH 6/6] staging: sm750fb: Spinlock and unlock in the same block

[Dan Carpenter]

On Wed, Mar 11, 2015 at 01:28:45AM +0000, Lorenzo Stoakes wrote:
> -static inline void myspin_lock(spinlock_t * sl){
> -     struct lynx_share * share;
> -     share = container_of(sl,struct lynx_share,slock);
> -     if(share->dual){
> -             spin_lock(sl);
> -     }
> -}

Yes, good.  We all hate locking wrappers but these are worse than
normal.

> +     /* if not use spin_lock,system will die if user load driver
> +      * and immediatly unload driver frequently (dual)*/
> +     if (share->dual) {
> +             spin_lock(&share->slock);
> +             share->accel.de_fillrect(&share->accel,
> +                                     base,pitch,Bpp,
> +                                     region->dx,region->dy,
> +                                     region->width,region->height,
> +                                     color,rop);
> +             spin_unlock(&share->slock);
> +     } else
> +             share->accel.de_fillrect(&share->accel,
> +                                     base,pitch,Bpp,
> +                                     region->dx,region->dy,
> +                                     region->width,region->height,
> +                                     color,rop);
>  }

No.  You've made the code uglier to work around Sparse stupidness.  Also
the braces are not according to kernel style.

        if (share->dual)
                spin_lock(&share->slock);

        share->accel.de_fillrect(&share->accel,
                                 base,pitch,Bpp,
                                 region->dx,region->dy,
                                 region->width,region->height,
                                 color,rop);

        if (share->dual)
                spin_unlock(&share->slock);

Sparse will still complain but no one cares.

regards,
_______________________________________________
devel mailing list
de...@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel