[Ocfs2-test-devel] [PATCH 37/59] reflink-test: fix mutiple reflink test race

[Eric Ren]

Hi Junxiao,

Looks good!

Reviewed-by: Eric Ren <zren at suse.com>

Thanks,
Eric Ren


On Mon, Sep 14, 2015 at 10:44:23AM +0800, Junxiao Bi wrote: 
> Rank 2 and Rank 3 pick a random length of value to update xattr, Rank3 will
> read the xattr and do validation after update, if Rank 2 picks a xattr length
> larger than Rank 3's and update the xattr after Rank 3's update but before the
> validation, fgetxattr in Rank 3 will return ERANGE error. Set "xattr_value_sz"
> to the max value to fix it.
> 
> Signed-off-by: Junxiao Bi <junxiao.bi at oracle.com>
> ---
>  programs/reflink_tests/multi_reflink_test.c |    1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/programs/reflink_tests/multi_reflink_test.c b/programs/reflink_tests/multi_reflink_test.c
> index 45c96fe..bd980c6 100755
> --- a/programs/reflink_tests/multi_reflink_test.c
> +++ b/programs/reflink_tests/multi_reflink_test.c
> @@ -1116,6 +1116,7 @@ static int do_xattr_data_cows(char *ref_pfx, unsigned long iter, int ea_nums)
>  
>  			if (xattr_value_sz > xattr_name_sz + 50) {
>  
> +				xattr_value_sz = XATTR_VALUE_MAX_SZ;
>  				ret = read_ea(NORMAL, fd);
>  				if (ret < 0)
>  					goto bail;
> -- 
> 1.7.9.5
> 
> 
> _______________________________________________
> Ocfs2-test-devel mailing list
> Ocfs2-test-devel at oss.oracle.com
> https://oss.oracle.com/mailman/listinfo/ocfs2-test-devel
>