On 08/02/2018 02:05 PM, Nir Soffer wrote:
> fallocate(FALLOC_FL_ZERO_RANGE) is supportd for block devices with
> modern kernel, but when it is not, we fall back to manual zeroing.
>
> Check if the underlying file is a block device when opening the file,
> and fall back to ioctl(BLKZEROOUT) for aligned zero requests for a
> block device.
>
> +++ b/plugins/file/file.c
> @@ -45,6 +45,7 @@
>   
>   #if defined(__linux__) && !defined(FALLOC_FL_PUNCH_HOLE)
>   #include <linux/falloc.h>   /* For FALLOC_FL_*, glibc < 2.18 */
> +#include <linux/fs.h>       /* For BLKZEROOUT */

Will this pick up BLKZEROOUT in all cases where it is needed?  Or do we
need to relax the !defined(FALLOC_FL_PUNCH_HOLE), and just blindly
include both of these headers for all Linux compilations?

> +static bool
> +is_aligned(struct handle *h, uint64_t n)
> +{
> +  return n % h->sector_size == 0;

Since we know (but the compiler doesn't) that sector_size is a power of
2, it is slightly faster to use bitwise math:
  return !(n & (h->sector_size - 1))

> +#ifdef BLKSSZGET
> +  if (h->is_block_device) {
> +    if (ioctl (h->fd, BLKSSZGET, &h->sector_size)) {
> +      nbdkit_error ("ioctl(BLKSSZGET): %s: %m", filename);
> +      free (h);
> +      return NULL;

If the ioctl() fails, would it be better to just fall back...

> +    }
> +  }
> +#else
> +  h->sector_size = 4096;  /* Safe guess */

...to the safe guess, instead of giving up entirely?  (Might matter on a
system with newer headers that have the macro, but where the kernel does
not support the ioctl).

> @@ -329,6 +361,20 @@ file_zero (void *handle, uint32_t count, uint64_t offset, int may_trim)
>     }
>   #endif
>   
> +#ifdef BLKZEROOUT
> +  /* For aligned range and block devices, we can use BLKZEROOUT. */
> +  if (h->is_block_device && is_aligned (h, offset) && is_aligned (h, count)) {

Since alignment is a power of 2, you can compress this as:

if (h->is_block_device && is_aligned (h, offset | count)) {

> +    uint64_t range[2] = {offset, count};
> +
> +    r = ioctl (h->fd, BLKZEROOUT, &range);
> +    if (r == 0)
> +      return r;
> +
> +    nbdkit_error ("zero: %m");
> +    return r;

Are we sure that treating ALL errors as fatal is worthwhile, or should
we still attempt to trigger a fall back to writing?

> +  }
> +#endif
> +
>     /* Trigger a fall back to writing */
>     errno = EOPNOTSUPP;
>     return r;
>

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