Received-SPF: pass (google.com: domain of linux-kernel+bounces-149535-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:40f1:3f00::1 as permitted sender) client-ip=2604:1380:40f1:3f00::1;
Date: Thu, 18 Apr 2024 11:13:19 +0530
From: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
To: Niklas Cassel <cassel@kernel.org>
Cc: Krzysztof =?utf-8?Q?Wilczy=C5=84ski?= <kw@linux.com>,
	Kishon Vijay Abraham I <kishon@kernel.org>,
	Bjorn Helgaas <bhelgaas@google.com>, linux-pci@vger.kernel.org,
	linux-kernel@vger.kernel.org,
	Dan Carpenter <dan.carpenter@linaro.org>
Subject: Re: [PATCH] PCI: endpoint: pci-epf-test: Make use of cached
 'epc_features' in pci_epf_test_core_init()
Message-ID: <20240418054319.GB2861@thinkpad>
References: <20240417-pci-epf-test-fix-v1-1-653c911d1faa@linaro.org>
 <ZiALuYlshLmwLhvu@ryzen>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <ZiALuYlshLmwLhvu@ryzen>

On Wed, Apr 17, 2024 at 07:49:45PM +0200, Niklas Cassel wrote:
> On Wed, Apr 17, 2024 at 10:47:25PM +0530, Manivannan Sadhasivam wrote:
> > Instead of getting the epc_features from pci_epc_get_features() API, use
> > the cached pci_epf_test::epc_features value to avoid the NULL check. Since
> > the NULL check is already performed in pci_epf_test_bind(), having one more
> > check in pci_epf_test_core_init() is redundant and it is not possible to
> > hit the NULL pointer dereference. This also leads to the following smatch
> > warning:
> > 
> > drivers/pci/endpoint/functions/pci-epf-test.c:784 pci_epf_test_core_init()
> > error: we previously assumed 'epc_features' could be null (see line 747)
> > 
> > Reported-by: Dan Carpenter <dan.carpenter@linaro.org>
> > Closes: https://lore.kernel.org/linux-pci/024b5826-7180-4076-ae08-57d2584cca3f@moroto.mountain/
> > Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
> 
> I think you forgot:
> Fixes: a01e7214bef9 ("PCI: endpoint: Remove "core_init_notifier" flag")
> 

No, that's not the correct fixes tag I suppose. This redudant check is
introduced by commit, 5e50ee27d4a5 ("PCI: pci-epf-test: Add support to defer
core initialization") and this commit removes the redundant check (fixing smatch
warning is a side effect). So if the fixes tag needs to be added, then this
commit should be referenced.

> 
> > ---
> >  drivers/pci/endpoint/functions/pci-epf-test.c | 9 ++++-----
> >  1 file changed, 4 insertions(+), 5 deletions(-)
> > 
> > diff --git a/drivers/pci/endpoint/functions/pci-epf-test.c b/drivers/pci/endpoint/functions/pci-epf-test.c
> > index 977fb79c1567..0d28f413cb07 100644
> > --- a/drivers/pci/endpoint/functions/pci-epf-test.c
> > +++ b/drivers/pci/endpoint/functions/pci-epf-test.c
> > @@ -743,11 +743,10 @@ static int pci_epf_test_core_init(struct pci_epf *epf)
> >  	bool msi_capable = true;
> >  	int ret;
> >  
> > -	epc_features = pci_epc_get_features(epc, epf->func_no, epf->vfunc_no);
> > -	if (epc_features) {
> > -		msix_capable = epc_features->msix_capable;
> > -		msi_capable = epc_features->msi_capable;
> > -	}
> > +	epc_features = epf_test->epc_features;
> 
> How about:
> 
> index 977fb79c1567..4d6105c07ac0 100644
> --- a/drivers/pci/endpoint/functions/pci-epf-test.c
> +++ b/drivers/pci/endpoint/functions/pci-epf-test.c
> @@ -735,7 +735,7 @@ static int pci_epf_test_core_init(struct pci_epf *epf)
>  {
>         struct pci_epf_test *epf_test = epf_get_drvdata(epf);
>         struct pci_epf_header *header = epf->header;
> -       const struct pci_epc_features *epc_features;
> +       const struct pci_epc_features *epc_features = epf_test->epc_features;
>         struct pci_epc *epc = epf->epc;
>         struct device *dev = &epf->dev;
>         bool linkup_notifier = false;
> @@ -743,12 +743,6 @@ static int pci_epf_test_core_init(struct pci_epf *epf)
>         bool msi_capable = true;
>         int ret;
>  
> -       epc_features = pci_epc_get_features(epc, epf->func_no, epf->vfunc_no);
> -       if (epc_features) {
> -               msix_capable = epc_features->msix_capable;
> -               msi_capable = epc_features->msi_capable;
> -       }
> -
>         if (epf->vfunc_no <= 1) {
>                 ret = pci_epc_write_header(epc, epf->func_no, epf->vfunc_no, header);
>                 if (ret) {
> @@ -761,6 +755,7 @@ static int pci_epf_test_core_init(struct pci_epf *epf)
>         if (ret)
>                 return ret;
>  
> +       msi_capable = epc_features->msi_capable;
>         if (msi_capable) {
>                 ret = pci_epc_set_msi(epc, epf->func_no, epf->vfunc_no,
>                                       epf->msi_interrupts);
> @@ -770,6 +765,7 @@ static int pci_epf_test_core_init(struct pci_epf *epf)
>                 }
>         }
>  
> +       msix_capable = epc_features->msix_capable;
>         if (msix_capable) {
>                 ret = pci_epc_set_msix(epc, epf->func_no, epf->vfunc_no,
>                                        epf->msix_interrupts,
> @@ -814,11 +810,9 @@ static int pci_epf_test_alloc_space(struct pci_epf *epf)
>         void *base;
>         enum pci_barno test_reg_bar = epf_test->test_reg_bar;
>         enum pci_barno bar;
> -       const struct pci_epc_features *epc_features;
> +       const struct pci_epc_features *epc_features = epf_test->epc_features;
>         size_t test_reg_size;
>  
> -       epc_features = epf_test->epc_features;
> -
>         test_reg_bar_size = ALIGN(sizeof(struct pci_epf_test_reg), 128);
>  
>         msix_capable = epc_features->msix_capable;
> 
> 
> Instead?
> 
> That way, we assign msi_capable/msix_capable just before the if-statement
> where it is used. (Which matches how we already assign msix_capable just
> before the if-statement in pci_epf_test_alloc_space().)
> 

Ok, if we go with this pattern, then pci_epf_test_set_bar() also needs to be
updated.

- Mani

-- 
மணிவண்ணன் சதாசிவம்