Re: [PATCH v11 07/23] CXL/PCI: Move CXL DVSEC definitions into uapi/linux/pci_regs.h

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Wed, 27 Aug 2025 16:51:15 +0200
Lukas Wunner <lukas@xxxxxxxxx> wrote:

> On Tue, Aug 26, 2025 at 08:35:22PM -0500, Terry Bowman wrote:
> > The CXL DVSECs are currently defined in cxl/core/cxlpci.h. These are not
> > accessible to other subsystems.
> > 
> > Change DVSEC name formatting to follow the existing PCI format in
> > pci_regs.h. The current format uses CXL_DVSEC_XYZ. Change to be PCI_DVSEC_CXL_XYZ.  
> [...]
> > +++ b/include/uapi/linux/pci_regs.h
> > @@ -1225,9 +1225,61 @@
> >  /* Deprecated old name, replaced with PCI_DOE_DATA_OBJECT_DISC_RSP_3_TYPE */
> >  #define PCI_DOE_DATA_OBJECT_DISC_RSP_3_PROTOCOL		PCI_DOE_DATA_OBJECT_DISC_RSP_3_TYPE
> >  
> > -/* Compute Express Link (CXL r3.1, sec 8.1.5) */
> > -#define PCI_DVSEC_CXL_PORT				3
> > -#define PCI_DVSEC_CXL_PORT_CTL				0x0c
> > -#define PCI_DVSEC_CXL_PORT_CTL_UNMASK_SBR		0x00000001
> > +/* Compute Express Link (CXL r3.2, sec 8.1)
> > + *
> > + * Note that CXL DVSEC id 3 and 7 to be ignored when the CXL link state
> > + * is "disconnected" (CXL r3.2, sec 9.12.3). Re-enumerate these
> > + * registers on downstream link-up events.
> > + */
> > +
> > +#define PCI_DVSEC_HEADER1_LENGTH_MASK  GENMASK(31, 20)
> > +
> > +/* CXL 3.2 8.1.3: PCIe DVSEC for CXL Device */
> > +#define PCI_DVSEC_CXL_DEVICE					0
> > +#define	  PCI_DVSEC_CXL_CAP_OFFSET	       0xA
> > +#define	    PCI_DVSEC_CXL_MEM_CAPABLE	       BIT(2)
> > +#define	    PCI_DVSEC_CXL_HDM_COUNT_MASK       GENMASK(5, 4)
> > +#define	  PCI_DVSEC_CXL_CTRL_OFFSET	       0xC
> > +#define	    PCI_DVSEC_CXL_MEM_ENABLE	       BIT(2)
> > +#define	  PCI_DVSEC_CXL_RANGE_SIZE_HIGH(i)     (0x18 + (i * 0x10))
> > +#define	  PCI_DVSEC_CXL_RANGE_SIZE_LOW(i)      (0x1C + (i * 0x10))
> > +#define	    PCI_DVSEC_CXL_MEM_INFO_VALID       BIT(0)
> > +#define	    PCI_DVSEC_CXL_MEM_ACTIVE	       BIT(1)
> > +#define	    PCI_DVSEC_CXL_MEM_SIZE_LOW_MASK    GENMASK(31, 28)
> > +#define	  PCI_DVSEC_CXL_RANGE_BASE_HIGH(i)     (0x20 + (i * 0x10))
> > +#define	  PCI_DVSEC_CXL_RANGE_BASE_LOW(i)      (0x24 + (i * 0x10))
> > +#define	    PCI_DVSEC_CXL_MEM_BASE_LOW_MASK    GENMASK(31, 28)  
> 
> Is it legal to use BIT() in a uapi header?
> 
> We've only allowed use of GENMASK() in uapi headers since 2023 with
> commit 3c7a8e190bc5 ("uapi: introduce uapi-friendly macros for GENMASK").
> 

I was curious, so went looking at that series.
There was a second patch that has what we need here
https://lore.kernel.org/all/20240131225010.2872733-3-pbonzini@xxxxxxxxxx/

So upshot is we should use _BITUL() for user space headers defined in
uapi/linux/const.h, also should be using  __GENMASK() not GENMASK()
in uapi/cxl/features.h

Anyhow fancy fixing up the existing cxl headers?  I'll get to it maybe but
will be a while as have a massive review backlog.

Jonathan



> But there is no uapi header in the kernel tree defining BIT().
> 
> I note that include/uapi/cxl/features.h has plenty of occurrences of BIT()
> since commit 9b8e73cdb141 ("cxl: Move cxl feature command structs to user
> header"), which went into v6.15.
> 
> ndctl contains a bitmap.h header defining BIT(), I guess that's why this
> wasn't perceived as a problem so far:
> https://github.com/pmem/ndctl/raw/main/util/bitmap.h
> 
> However existing user space applications including <linux/pci_regs.h>
> may not have a BIT() definition and I suspect your change above will
> break the build of those applications.
> 
> Thanks,
> 
> Lukas
> 





[Index of Archives]     [DMA Engine]     [Linux Coverity]     [Linux USB]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [Greybus]

  Powered by Linux