[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH for-5.1 2/2] file-posix: Allow byte-aligned O_DIRECT with NFS
From: |
Nir Soffer |
Subject: |
Re: [PATCH for-5.1 2/2] file-posix: Allow byte-aligned O_DIRECT with NFS |
Date: |
Mon, 13 Jul 2020 19:29:26 +0300 |
On Fri, Jul 10, 2020 at 5:22 PM Kevin Wolf <kwolf@redhat.com> wrote:
>
> Since commit a6b257a08e3 ('file-posix: Handle undetectable alignment'),
> we assume that if we open a file with O_DIRECT and alignment probing
> returns 1, we just couldn't find out the real alignment requirement
> because some filesystems make the requirement only for allocated blocks.
> In this case, a safe default of 4k is used.
>
> This is too strict NFS, which does actually allow byte-aligned requests
> even with O_DIRECT. Because we can't distinguish both cases with generic
> code, let's just look at the file system magic and disable
> s->needs_alignment for NFS. This way, O_DIRECT can still be used on NFS
> for images that are not aligned to 4k.
>
> Signed-off-by: Kevin Wolf <kwolf@redhat.com>
> ---
> block/file-posix.c | 26 +++++++++++++++++++++++++-
> 1 file changed, 25 insertions(+), 1 deletion(-)
>
> diff --git a/block/file-posix.c b/block/file-posix.c
> index 0c4e07c415..4e9dac461b 100644
> --- a/block/file-posix.c
> +++ b/block/file-posix.c
> @@ -62,10 +62,12 @@
> #include <sys/ioctl.h>
> #include <sys/param.h>
> #include <sys/syscall.h>
> +#include <sys/vfs.h>
> #include <linux/cdrom.h>
> #include <linux/fd.h>
> #include <linux/fs.h>
> #include <linux/hdreg.h>
> +#include <linux/magic.h>
> #include <scsi/sg.h>
> #ifdef __s390__
> #include <asm/dasd.h>
> @@ -300,6 +302,28 @@ static int probe_physical_blocksize(int fd, unsigned int
> *blk_size)
> #endif
> }
>
> +/*
> + * Returns true if no alignment restrictions are necessary even for files
> + * opened with O_DIRECT.
> + *
> + * raw_probe_alignment() probes the required alignment and assume that 1
> means
> + * the probing failed, so it falls back to a safe default of 4k. This can be
> + * avoided if we know that byte alignment is okay for the file.
> + */
> +static bool dio_byte_aligned(int fd)
> +{
> +#ifdef __linux__
> + struct statfs buf;
> + int ret;
> +
> + ret = fstatfs(fd, &buf);
> + if (ret == 0 && buf.f_type == NFS_SUPER_MAGIC) {
> + return true;
> + }
> +#endif
> + return false;
> +}
> +
> /* Check if read is allowed with given memory buffer and length.
> *
> * This function is used to check O_DIRECT memory buffer and request
> alignment.
> @@ -631,7 +655,7 @@ static int raw_open_common(BlockDriverState *bs, QDict
> *options,
>
> s->has_discard = true;
> s->has_write_zeroes = true;
> - if ((bs->open_flags & BDRV_O_NOCACHE) != 0) {
> + if ((bs->open_flags & BDRV_O_NOCACHE) != 0 && !dio_byte_aligned(s->fd)) {
> s->needs_alignment = true;
I did not know we have needs_alignment. Isn't this the same as using
request_alignment = 1?
For example we can check if we are on NFS and avoid the fallback to max_align:
if (!bs->bl.request_alignment) {
int i;
size_t align;
buf = qemu_memalign(max_align, max_align);
for (i = 0; i < ARRAY_SIZE(alignments); i++) {
align = alignments[i];
if (raw_is_io_aligned(fd, buf, align)) {
/* Fallback to safe value. */
bs->bl.request_alignment = (align != 1) ? align : max_align;
break;
}
}
qemu_vfree(buf);
}
After this we will have correct bl.request_alignment and buf_align.
Hopefully this will not break code expecting request_alignment >= 512.
Assuming that needs_alignment is well tested, this patch may be safer.
Nir
> }
>
> --
> 2.25.4
>