[LTP] [PATCH] fs_fill: Fix test when running on a 256 CPU+ machine

[Andrei Gherzan]

On 23/02/17 04:27PM, Petr Vorel wrote:
> Hi Andrei,
> 
> > The fs_fill test runs a fill test on all the supported filesystems. One
> > of them being vfat. This filesystem is configured dynamically or through
> > flags/arguments for its file allocation table type (12/16/32).
> 
> > The size of the test device (which is a loop-mounted fs) is 300MB. When not
> > instructed, mkfs will "automatically select between 12, 16 and 32 bit,
> > whatever fits better for the filesystem size"[1]. In the case of a 300Mb that
> > would end up as FAT16.
> Interesting. BTW we plan to change 300 MB to minimal filesystem which would fit
> to all existing tests (255 MB was for Btrfs, 300 MB was for XFS, but there might
> be minimal systems which can use vfat, ext4, ... with smaller resources, e.g.
> 16 MB for filesystem). Therefore I wonder what is minimal reasonable required
> size for vfat. i.e. what MB is required for FAT32? (I guess we don't want to
> check FAT12 or FAT16).

The minimum/maximum values depend on the sector size. I'm not sure about
F12 but for the other ones I think it was 32MB for 512b sectors.

> 
> > This is linked with another configuration that is the actual impact on
> > this issue: the maximum number of directories in the root of the
> > filesystem. FAT12 and FAT16 use a special system region: "Root Directory
> > Region". And by default (there is also an argument to configure this at
> > mkfs-time) this ends up being 256 when no custom arguments are provided.
> 
> > On the other hand, the test sets up the filesystem with a
> > "tst_ncpus_conf + 2" number of directories in the test filesystem's root
> > which can break the limit explained above on systems with a number of
> > CPUs that would amount above the limit.
> 
> > This change addresses this situation by using a subdirectory in the test
> > filesystem which is not subject to this limit.
> 
> > Signed-off-by: Andrei Gherzan <andrei.gherzan@canonical.com>
> > ---
> >  testcases/kernel/fs/fs_fill/fs_fill.c | 7 ++++++-
> >  1 file changed, 6 insertions(+), 1 deletion(-)
> 
> > diff --git a/testcases/kernel/fs/fs_fill/fs_fill.c b/testcases/kernel/fs/fs_fill/fs_fill.c
> > index 2027b6df1..dd8ad2935 100644
> > --- a/testcases/kernel/fs/fs_fill/fs_fill.c
> > +++ b/testcases/kernel/fs/fs_fill/fs_fill.c
> > @@ -18,6 +18,7 @@
> >  #include "tst_test.h"
> 
> >  #define MNTPOINT "mntpoint"
> > +#define THREADS_SUBDIR "subdir"
> 
> >  static volatile int run;
> >  static unsigned int nthreads;
> > @@ -99,9 +100,13 @@ static void setup(void)
> >  	nthreads = tst_ncpus_conf() + 2;
> >  	workers = SAFE_MALLOC(sizeof(struct worker) * nthreads);
> 
> > +	// Avoid creating the thread directories in the root of the filesystem
> > +	// to not hit the root entries limit on a  FAT16 filesystem.
> > +	SAFE_MKDIR(MNTPOINT "/" THREADS_SUBDIR, 0700);
> > +
> >  	for (i = 0; i < nthreads; i++) {
> >  		snprintf(workers[i].dir, sizeof(workers[i].dir),
> > -			 MNTPOINT "/thread%i", i + 1);
> > +			 MNTPOINT "/" THREADS_SUBDIR "/thread%i", i + 1);
> >  		SAFE_MKDIR(workers[i].dir, 0700);
> >  	}
> 
> Reviewed-by: Petr Vorel <pvorel@suse.cz>
> 
> If you don't mind, I suggest to merge slightly changed patch.

Happy with the changes below.

> 
> Kind regards,
> Petr
> 
> +++ testcases/kernel/fs/fs_fill/fs_fill.c
> @@ -18,6 +18,7 @@
>  #include "tst_test.h"
>  
>  #define MNTPOINT "mntpoint"
> +#define THREADS_DIR MNTPOINT "/subdir"
>  
>  static volatile int run;
>  static unsigned int nthreads;
> @@ -99,9 +100,15 @@ static void setup(void)
>  	nthreads = tst_ncpus_conf() + 2;
>  	workers = SAFE_MALLOC(sizeof(struct worker) * nthreads);
>  
> +	/*
> +	 * Avoid creating the thread directories in the root of the filesystem
> +	 * to not hit the root entries limit on a FAT16 filesystem.
> +	 */
> +	SAFE_MKDIR(THREADS_DIR, 0700);
> +
>  	for (i = 0; i < nthreads; i++) {
>  		snprintf(workers[i].dir, sizeof(workers[i].dir),
> -			 MNTPOINT "/thread%i", i + 1);
> +			 THREADS_DIR "/thread%i", i + 1);
>  		SAFE_MKDIR(workers[i].dir, 0700);
>  	}
>  

-- 
Andrei Gherzan