On Wed, Jun 19, 2019 at 11:28:57AM -0700, Allison Collins wrote:
> While investigating another mkfs bug, noticed that cfg->lsunit is sometimes
> left uninitialized when it should not. This is because calc_stripe_factors
> in some cases needs cfg->loginternal to be set first. This is done in
> validate_logdev. So move calc_stripe_factors below validate_logdev while
> parsing configs.
<grumble> The cfg in main() is not (in a manner easily detectable by
toolz) uninitialized, it's zero-initialized by default and we haven't
set cfg->loginternal correctly yet...
...what we really need here is enum { FALSE, TRUE, FILENOTFOUND } to
detect that we're using incorrect garbage data. :P
(Really, someone should take a closer look at whether or not there are
other places where we do things like this...)
Anyway, this does solve a problem, so
Reviewed-by: Darrick J. Wong <darrick.wong@xxxxxxxxxx>
--D
>
> Signed-off-by: Allison Collins <allison.henderson@xxxxxxxxxx>
> ---
> mkfs/xfs_mkfs.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/mkfs/xfs_mkfs.c b/mkfs/xfs_mkfs.c
> index ddb25ec..f4a5e4b 100644
> --- a/mkfs/xfs_mkfs.c
> +++ b/mkfs/xfs_mkfs.c
> @@ -3995,7 +3995,6 @@ main(
> cfg.rtblocks = calc_dev_size(cli.rtsize, &cfg, &ropts, R_SIZE, "rt");
>
> validate_rtextsize(&cfg, &cli, &ft);
> - calc_stripe_factors(&cfg, &cli, &ft);
>
> /*
> * Open and validate the device configurations
> @@ -4005,6 +4004,7 @@ main(
> validate_datadev(&cfg, &cli);
> validate_logdev(&cfg, &cli, &logfile);
> validate_rtdev(&cfg, &cli, &rtfile);
> + calc_stripe_factors(&cfg, &cli, &ft);
>
> /*
> * At this point when know exactly what size all the devices are,
> --
> 2.7.4
>
