Re: [PATCH] iotests: Fix unsupported_imgopts for refcount_bits

[Eric Blake <eblake@redhat.com>]

On 2/9/21 12:27 PM, Max Reitz wrote:
> Many _unsupported_imgopts lines for refcount_bits values use something
> like "refcount_bits=1[^0-9]" to forbid everything but "refcount_bits=1"
> (e.g. "refcount_bits=16" is allowed).
> 
> That does not work when $IMGOPTS does not have any entry past the
> refcount_bits option, which now became apparent with the "check" script
> rewrite.
> 
> Use \b instead of [^0-9] to check for a word boundary, which is what we
> really want.

\b is a Linux-ism (that is, glibc supports it, but BSD libc does not).

https://mail-index.netbsd.org/tech-userlevel/2012/12/02/msg006954.html


> 
> Signed-off-by: Max Reitz <mreitz@redhat.com>
> ---
> Reproducible with:
> $ ./check -qcow2 -o refcount_bits=1
> (The tests touched here should be skipped)
> 
> I don't know whether \b is portable.  I hope it is.
> (This is why I CC-ed you, Eric.)

No, it's not portable.  \> and [[:>:]] are other spellings for the same
task, equally non-portable.

> 
> Then again, it appears that nobody ever runs the iotests with
> refcount_bits=1 but me, and I do that on Linux.  So even if it isn't
> portable, it shouldn't be an issue in practice... O:)

What exactly is failing?  Is it merely a case of our python script
running the regex against "${unsupported_imgopts}" instead of
"${unsupported_imgsopts} " with an added trailing space to guarantee
that we have something to match against?

-- 
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3226
Virtualization:  qemu.org | libvirt.org