linux-kernel - Re: [PATCH v15 1/6] PCI: Clean up __pci_find_next_cap

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] [day] [month] [year] [list]

Message-ID: <fb7c14c5-0aaf-6f90-caac-71d6ce6f25ae@linux.intel.com>
Date: Wed, 20 Aug 2025 14:36:19 +0300 (EEST)
From: Ilpo Järvinen <ilpo.jarvinen@...ux.intel.com>
To: Gerd Bayer <gbayer@...ux.ibm.com>
cc: Hans Zhang <18255117159@....com>, lpieralisi@...nel.org, 
    kwilczynski@...nel.org, bhelgaas@...gle.com, helgaas@...nel.org, 
    jingoohan1@...il.com, mani@...nel.org, robh@...nel.org, 
    schnelle@...ux.ibm.com, Lukas Wunner <lukas@...ner.de>, arnd@...nel.org, 
    geert@...ux-m68k.org, linux-pci@...r.kernel.org, 
    LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v15 1/6] PCI: Clean up __pci_find_next_cap_ttl()
 readability

On Wed, 20 Aug 2025, Gerd Bayer wrote:

> On Wed, 2025-08-13 at 22:45 +0800, Hans Zhang wrote:
> > Refactor the __pci_find_next_cap_ttl() to improve code clarity:
> > - Replace magic number 0x40 with PCI_STD_HEADER_SIZEOF.
> > - Use ALIGN_DOWN() for position alignment instead of manual bitmask.
> > - Extract PCI capability fields via FIELD_GET() with standardized masks.
> > - Add necessary headers (linux/align.h).
> > 
> > No functional changes intended.
> > 
> > Signed-off-by: Hans Zhang <18255117159@....com>
> > Acked-by: Manivannan Sadhasivam <mani@...nel.org>
> > ---
> >  drivers/pci/pci.c             | 9 +++++----
> >  include/uapi/linux/pci_regs.h | 3 +++
> >  2 files changed, 8 insertions(+), 4 deletions(-)
> > 
> > diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> > index b0f4d98036cd..40a5c87d9a6b 100644
> > --- a/drivers/pci/pci.c
> > +++ b/drivers/pci/pci.c
> > @@ -9,6 +9,7 @@
> >   */
> >  
> >  #include <linux/acpi.h>
> > +#include <linux/align.h>
> >  #include <linux/kernel.h>
> >  #include <linux/delay.h>
> >  #include <linux/dmi.h>
> > @@ -432,17 +433,17 @@ static u8 __pci_find_next_cap_ttl(struct pci_bus *bus, unsigned int devfn,
> >  	pci_bus_read_config_byte(bus, devfn, pos, &pos);
> >  
> >  	while ((*ttl)--) {
> > -		if (pos < 0x40)
> > +		if (pos < PCI_STD_HEADER_SIZEOF)
> >  			break;
> > -		pos &= ~3;
> > +		pos = ALIGN_DOWN(pos, 4);
> >  		pci_bus_read_config_word(bus, devfn, pos, &ent);
> >  
> > -		id = ent & 0xff;
> > +		id = FIELD_GET(PCI_CAP_ID_MASK, ent);
> >  		if (id == 0xff)
> >  			break;
> >  		if (id == cap)
> >  			return pos;
> > -		pos = (ent >> 8);
> > +		pos = FIELD_GET(PCI_CAP_LIST_NEXT_MASK, ent);
> >  	}
> >  	return 0;
> >  }
> > diff --git a/include/uapi/linux/pci_regs.h b/include/uapi/linux/pci_regs.h
> > index f5b17745de60..1bba99b46227 100644
> > --- a/include/uapi/linux/pci_regs.h
> > +++ b/include/uapi/linux/pci_regs.h
> > @@ -207,6 +207,9 @@
> >  
> >  /* Capability lists */
> >  
> > +#define PCI_CAP_ID_MASK		0x00ff	/* Capability ID mask */
> > +#define PCI_CAP_LIST_NEXT_MASK	0xff00	/* Next Capability Pointer mask */
> > +
> >  #define PCI_CAP_LIST_ID		0	/* Capability ID */
> >  #define  PCI_CAP_ID_PM		0x01	/* Power Management */
> >  #define  PCI_CAP_ID_AGP		0x02	/* Accelerated Graphics Port */
> 
> Hi Hans,
> 
> I like your approach to replace the magic numbers here. If you went
> further to replace the single pci_bus_read_config_word() with two
> single-byte reads at the appropriate places - for CAP_ID and
> CAP_LIST_NEXT - you could even go with the already existing offset
> defines PCI_CAP_LIST_ID and PCI_CAP_LIST_NEXT from pci_regs.h.
>
> But that might be a more intricate change and involves more HW accesses
> than what it's worth.

Hi,

As you noted, it'll be less efficient so it's undesirable to split the 
read.

It's somewhat problematic that some of the defines in 
include/uapi/linux/pci_regs.h cannot be easily used efficiently with 
multi-byte reads leading to use of literals in code.

IMO, adding the multi-byte masks like this Hans' change is IMO the correct 
way to address it.

> So feel free to add my
> Reviewed-by: Gerd Bayer <gbayer@...ux.ibm.com>
> 
> Thanks,
> Gerd
> 

-- 
 i.