Re: [PATCH 09/10] generic/624: test multiple Merkle tree block sizes

[Eric Biggers <ebiggers@kernel.org>]

On Tue, Dec 20, 2022 at 02:56:04PM +0800, Zorro Lang wrote:
> > +# Always test FSV_BLOCK_SIZE.  Also test some other block sizes if they happen
> > +# to be supported.
> > +_fsv_scratch_begin_subtest "Testing FS_IOC_READ_VERITY_METADATA with block_size=FSV_BLOCK_SIZE"
> > +test_block_size $FSV_BLOCK_SIZE
> > +for block_size in 1024 4096 16384 65536; do
> > +	_fsv_scratch_begin_subtest "Testing FS_IOC_READ_VERITY_METADATA with block_size=$block_size"
> > +	if (( block_size == FSV_BLOCK_SIZE )); then
> > +		continue
> > +	fi
> > +	if ! _fsv_can_enable $fsv_file --block-size=$block_size; then
> > +		echo "block_size=$block_size is unsupported" >> $seqres.full
> > +		continue
> 
> If a block size isn't supported, e.g. 1024. Then this case trys to skip that
> test, but it'll break golden image, due to the .out file contains each line
> of:
>   Testing FS_IOC_READ_VERITY_METADATA with block_size=1024/4096/16384/65536
> 
> Do you expect that failure, or we shouldn't fail on that?

Actually it doesn't fail, since "Testing FS_IOC_READ_VERITY_METADATA with
block_size=$block_size" is printed unconditionally, and
"block_size=$block_size is unsupported" is only printed to $seqres.full.

To avoid this confusion, how about I change "Testing FS_IOC_READ_VERITY_METADATA
with block_size=$block_size" to "Testing block_size=$block_size if supported"?
Or do you have another suggestion?

- Eric