Re: [PATCH] 9p/xen: Use flexible array for data rings

[Dominique Martinet]

Jürgen Groß wrote on Wed, May 27, 2026 at 07:39:07AM +0200:
> On 19.05.26 03:57, Rosen Penev wrote:
> > Store the fixed set of Xen 9p data rings in the frontend private
> > allocation instead of allocating a separate rings array.
> > 
> > This keeps the data ring storage tied to the frontend lifetime and
> > simplifies the allocation and cleanup paths.
> > 
> > Assisted-by: Codex:GPT-5.5
> > Signed-off-by: Rosen Penev <rosenp@xxxxxxxxx>
> > ---
> >   net/9p/trans_xen.c | 68 ++++++++++++++++++++--------------------------
> >   1 file changed, 30 insertions(+), 38 deletions(-)
> > 
> > diff --git a/net/9p/trans_xen.c b/net/9p/trans_xen.c
> > index f9fb2db7a066..5a110d71d18c 100644
> > --- a/net/9p/trans_xen.c
> > +++ b/net/9p/trans_xen.c
> > @@ -55,7 +55,7 @@ struct xen_9pfs_front_priv {
> >   	char *tag;
> >   	struct p9_client *client;
> > -	struct xen_9pfs_dataring *rings;
> > +	struct xen_9pfs_dataring rings[];
> 
> Any reason not to use rings[XEN_9PFS_NUM_RINGS] instead of a flex array?

Rosen, do you intend to resend with this fixed number?

Given that's just a two line diff from your patch (just changing the
type and keeping kzalloc_obj instead of kzalloc_flex) I can do it myself
but I also don't have any direct merit from it, so if I don't hear back
I'll forget about this
(unless Jürgen says he actually read the whole diff and it's worth
taking for simplification, I don't have time...)

Thanks,
-- 
Dominique Martinet | Asmadeus