From mboxrd@z Thu Jan 1 00:00:00 1970 From: Paul Mackerras Date: Wed, 26 Sep 2018 11:21:06 +0000 Subject: Re: [RFC PATCH 12/32] powerpc: Provide dummy versions of plpar_hcall functions when !CONFIG_PPC_PSER Message-Id: <20180926112106.GD29799@fergus> List-Id: References: <1537524123-9578-13-git-send-email-paulus@ozlabs.org> In-Reply-To: <1537524123-9578-13-git-send-email-paulus@ozlabs.org> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: kvm-ppc@vger.kernel.org On Wed, Sep 26, 2018 at 01:30:40PM +1000, David Gibson wrote: > On Fri, Sep 21, 2018 at 08:01:43PM +1000, Paul Mackerras wrote: > > This will reduce the number of instances where we require ifdefs in > > code which needs to call a hypercall on a pseries platform or do > > something different on a powernv platform. > > > > Signed-off-by: Paul Mackerras > > Wouldn't it make more sense for these to BUG_ON() or at least > WARN_ON() before returning H_FUNCTION, since calling a hypercall on a > non pseries platform is definitely a kernel bug? Not BUG_ON, since at this level we have no indication that the kernel can't continue executing without whatever it was that it expected the hcall to do. I could come at a WARN_ON (or WARN_ON_ONCE) -- I'll talk to mpe and see what he thinks. Paul.