Re: [PATCH] cxgb4: fix TLS dependencies again

[Arnd Bergmann]

On Sun, Jan 3, 2021 at 5:51 PM James Bottomley <jejb@xxxxxxxxxxxxx> wrote:
> On Sun, 2021-01-03 at 15:01 +0100, Arnd Bergmann wrote:
> > From: Arnd Bergmann <arnd@xxxxxxxx>
> >
> > A previous patch tried to avoid a build failure, but missed one case
> > that Kconfig warns about:
> >
> > WARNING: unmet direct dependencies detected for CHELSIO_T4
> >   Depends on [m]: NETDEVICES [=y] && ETHERNET [=y] &&
> > NET_VENDOR_CHELSIO [=y] && PCI [=y] && (IPV6 [=y] || IPV6 [=y]=n) &&
> > (TLS [=m] || TLS [=m]=n)
> >   Selected by [y]:
> >   - SCSI_CXGB4_ISCSI [=y] && SCSI_LOWLEVEL [=y] && SCSI [=y] && PCI
> > [=y] && INET [=y] && (IPV6 [=y] || IPV6 [=y]=n) && ETHERNET [=y]
> > x86_64-linux-ld: drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.o: in
> > function `cxgb_select_queue':
> > cxgb4_main.c:(.text+0xf5df): undefined reference to
> > `tls_validate_xmit_skb'
> >
> > When any of the dependencies of CHELSIO_T4 are not met, then
> > SCSI_CXGB4_ISCSI must not 'select' it either.
> >
> > Fix it by mirroring the network driver dependencies on the iscsi
> > driver. A more invasive but also more reliable alternative would
> > be to use 'depends on CHELSIO_T4' instead.
> >
> > Fixes: 659fbdcf2f14 ("cxgb4: Fix build failure when CONFIG_TLS=m")
> > Signed-off-by: Arnd Bergmann <arnd@xxxxxxxx>
> > ---
> >  drivers/scsi/cxgbi/cxgb4i/Kconfig | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/drivers/scsi/cxgbi/cxgb4i/Kconfig
> > b/drivers/scsi/cxgbi/cxgb4i/Kconfig
> > index 8b0deece9758..2af88a55fbca 100644
> > --- a/drivers/scsi/cxgbi/cxgb4i/Kconfig
> > +++ b/drivers/scsi/cxgbi/cxgb4i/Kconfig
> > @@ -1,7 +1,7 @@
> >  # SPDX-License-Identifier: GPL-2.0-only
> >  config SCSI_CXGB4_ISCSI
> >       tristate "Chelsio T4 iSCSI support"
> > -     depends on PCI && INET && (IPV6 || IPV6=n)
> > +     depends on PCI && INET && (IPV6 || IPV6=n) && (TLS || TLS=n)
> >       depends on THERMAL || !THERMAL
> >       depends on ETHERNET
> >       depends on TLS || TLS=n
>
> I thought all separated depends statements were the equivalent of && in
> a single statement.  If so, how does repeating TLS || TLS=n twice fix
> the problem?  This sounds more like we have some sort of bug in our
> Kconfig apparatus.

No, just a mistake on my end. I had tried v5.10-rc1 to make sure the
patch was still required, but then rebased on top of torvalds/master
(close to -rc2) before sending it out, and this contains Randy's version
of the same patch, cb5253198f10 ("scsi: cxgb4i: Fix TLS dependency").

       Arnd