linux-kernel - Re: [PATCH] PCI: iproc: Set all 24 bits of PCI class code

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Wed, 5 Jan 2022 19:13:06 +0100
From:   Pali Rohár <pali@...nel.org>
To:     Ray Jui <ray.jui@...adcom.com>
Cc:     Roman Bacik <roman.bacik@...adcom.com>,
        Lorenzo Pieralisi <lorenzo.pieralisi@....com>,
        Rob Herring <robh@...nel.org>,
        Krzysztof Wilczyński <kw@...ux.com>,
        Bjorn Helgaas <bhelgaas@...gle.com>,
        bcm-kernel-feedback-list@...adcom.com, linux-pci@...r.kernel.org,
        linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] PCI: iproc: Set all 24 bits of PCI class code

Hello!

On Wednesday 05 January 2022 09:51:48 Ray Jui wrote:
> Hi Pali,
> 
> On 1/5/2022 1:35 AM, Pali Rohár wrote:
> > Register 0x43c in its low 24 bits contains PCI class code.
> > 
> > Update code to set all 24 bits of PCI class code and not only upper 16 bits
> > of PCI class code.
> > 
> > Use a new macro PCI_CLASS_BRIDGE_PCI_NORMAL which represents whole 24 bits
> > of normal PCI bridge class.
> > 
> > Signed-off-by: Pali Rohár <pali@...nel.org>
> > 
> > ---
> > Roman helped me with this change and confirmed that class code is stored
> > really in bits [23:0] of custom register 0x43c (normally class code is
> > stored in bits [31:8] of pci register 0x08).
> > 
> > This patch depends on patch which adds PCI_CLASS_BRIDGE_PCI_NORMAL macro:
> > https://lore.kernel.org/linux-pci/20211220145140.31898-1-pali@kernel.org/
> > ---
> >  drivers/pci/controller/pcie-iproc.c | 9 ++++-----
> >  1 file changed, 4 insertions(+), 5 deletions(-)
> > 
> > diff --git a/drivers/pci/controller/pcie-iproc.c b/drivers/pci/controller/pcie-iproc.c
> > index 3df4ab209253..2519201b0e51 100644
> > --- a/drivers/pci/controller/pcie-iproc.c
> > +++ b/drivers/pci/controller/pcie-iproc.c
> > @@ -789,14 +789,13 @@ static int iproc_pcie_check_link(struct iproc_pcie *pcie)
> >  		return -EFAULT;
> >  	}
> >  
> > -	/* force class to PCI_CLASS_BRIDGE_PCI (0x0604) */
> > +	/* force class to PCI_CLASS_BRIDGE_PCI_NORMAL (0x060400) */
> >  #define PCI_BRIDGE_CTRL_REG_OFFSET	0x43c
> > -#define PCI_CLASS_BRIDGE_MASK		0xffff00
> > -#define PCI_CLASS_BRIDGE_SHIFT		8
> > +#define PCI_BRIDGE_CTRL_REG_CLASS_MASK	0xffffff
> >  	iproc_pci_raw_config_read32(pcie, 0, PCI_BRIDGE_CTRL_REG_OFFSET,
> >  				    4, &class);
> > -	class &= ~PCI_CLASS_BRIDGE_MASK;
> > -	class |= (PCI_CLASS_BRIDGE_PCI << PCI_CLASS_BRIDGE_SHIFT);
> > +	class &= ~PCI_BRIDGE_CTRL_REG_CLASS_MASK;
> > +	class |= PCI_CLASS_BRIDGE_PCI_NORMAL;
> >  	iproc_pci_raw_config_write32(pcie, 0, PCI_BRIDGE_CTRL_REG_OFFSET,
> >  				     4, class);
> >  
> 
> I have two comments:
> 
> 1. You do not seem to generate the email list using the
> get_maintainer.pl script, so the two maintainers for Broadcom ARM
> architecture (Ray Jui and Scott Branden) are left out.

Ou, sorry for that! I have generated this patch for U-Boot and Linux
kernel and probably mixed or forgot to include correct recipients for
correct project.

> 2. I suppose 'PCI_CLASS_BRIDGE_PCI_NORMAL' is defined in some common PCI
> header in a separate patch as described in the commit message. Then how
> come these patches are not constructed with a patch series?

Yes, PCI_CLASS_BRIDGE_PCI_NORMAL is a new constant for common pci header
file defined in patch linked in commit message.
https://lore.kernel.org/linux-pci/20211220145140.31898-1-pali@kernel.org/

Originally I included this change in v1 of linked patch in December but
I realized that it does not match standard PCI config space (different
offset 0x43c vs 0x08 and also different shift 0x8 vs 0x0) and probably
there is something either incorrect or really non-standard. So later in
December I dropped iproc_pcie_check_link() change in v2 of the linked
patch where is introduced PCI_CLASS_BRIDGE_PCI_NORMAL and now sent new
change for iproc_pcie_check_link() separately.

Technically, linked patch in commit message is just extracting code into
the common macros without any functional changed. But change in this
iproc_pcie_check_link() has also functional change as now also lower 8
bits of class code are changed. So in my opinion this patch should be
really separate of linked patch.

I hope that Lorenzo and Bjorn take patches in correct order...

> Other than, the change itself is exactly what I sent to Roman and looks
> good to me. Thanks.
> 
> Acked-by: Ray Jui <ray.jui@...adcom.com>

Perfect!