Re: [Qemu-devel] [PATCH 2/3] block: set bs->read_only before .bdrv_open()

[Kevin Wolf]

Am 27.10.2011 11:54, schrieb Stefan Hajnoczi:
> Several block drivers set bs->read_only in .bdrv_open() but
> block.c:bdrv_open_common() clobbers its value.  Additionally, QED uses
> bdrv_is_read_only() in .bdrv_open() to decide whether to perform
> consistency checks.
> 
> The correct ordering is to initialize bs->read_only from the open flags
> before calling .bdrv_open().  This way block drivers can override it if
> necessary and can use bdrv_is_read_only() in .bdrv_open().
> 
> Signed-off-by: Stefan Hajnoczi <address@hidden>
> ---
>  block.c |    4 ++--
>  1 files changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/block.c b/block.c
> index 70aab63..3207e99 100644
> --- a/block.c
> +++ b/block.c
> @@ -500,6 +500,8 @@ static int bdrv_open_common(BlockDriverState *bs, const 
> char *filename,
>          open_flags |= BDRV_O_RDWR;
>      }

Not directly related, but the context made me wonder when we're making a
BlockkDriverState writeable unconditionally. This is the full context:


    /*
     * Snapshots should be writable.
     */
    if (bs->is_temporary) {
        open_flags |= BDRV_O_RDWR;
    }

Does anyone understand what the point of this is? If the user requested
read-only, he certainly wants to have read-only, even if he specified
-snapshot as well.

> +    bs->keep_read_only = bs->read_only = !(open_flags & BDRV_O_RDWR);
> +
>      /* Open the image, either directly or using a protocol */
>      if (drv->bdrv_file_open) {
>          ret = drv->bdrv_file_open(bs, filename, open_flags);
> @@ -514,8 +516,6 @@ static int bdrv_open_common(BlockDriverState *bs, const 
> char *filename,
>          goto free_and_fail;
>      }
>  
> -    bs->keep_read_only = bs->read_only = !(open_flags & BDRV_O_RDWR);
> -

The assignment was already at the new place before 4dca4b6. Not sure if
there was any real reason for moving it, though.

Kevin