1 Accessing PCI device resources through sysfs
2 --------------------------------------------
4 sysfs, usually mounted at /sys, provides access to PCI resources on platforms
5 that support it. For example, a given bus might look like this:
7 /sys/devices/pci0000:17
21 | |-- subsystem_device
22 | |-- subsystem_vendor
26 The topmost element describes the PCI domain and bus number. In this case,
27 the domain number is 0000 and the bus number is 17 (both values are in hex).
28 This bus contains a single function device in slot 0. The domain and bus
29 numbers are reproduced for convenience. Under the device directory are several
30 files, each with their own function.
34 class PCI class (ascii, ro)
35 config PCI config space (binary, rw)
36 device PCI device (ascii, ro)
37 enable Whether the device is enabled (ascii, rw)
38 irq IRQ number (ascii, ro)
39 local_cpus nearby CPU mask (cpumask, ro)
40 remove remove device from kernel's list (ascii, wo)
41 resource PCI resource host addresses (ascii, ro)
42 resource0..N PCI resource N, if present (binary, mmap)
43 resource0_wc..N_wc PCI WC map resource N, if prefetchable (binary, mmap)
44 rom PCI ROM resource, if present (binary, ro)
45 subsystem_device PCI subsystem device (ascii, ro)
46 subsystem_vendor PCI subsystem vendor (ascii, ro)
47 vendor PCI vendor (ascii, ro)
50 rw - file is readable and writable
52 mmap - file is mmapable
53 ascii - file contains ascii text
54 binary - file contains binary data
55 cpumask - file contains a cpumask type
57 The read only files are informational, writes to them will be ignored, with
58 the exception of the 'rom' file. Writable files can be used to perform
59 actions on the device (e.g. changing config space, detaching a device).
60 mmapable files are available via an mmap of the file at offset 0 and can be
61 used to do actual device programming from userspace. Note that some platforms
62 don't support mmapping of certain resources, so be sure to check the return
63 value from any attempted mmap.
65 The 'enable' file provides a counter that indicates how many times the device
66 has been enabled. If the 'enable' file currently returns '4', and a '1' is
67 echoed into it, it will then return '5'. Echoing a '0' into it will decrease
68 the count. Even when it returns to 0, though, some of the initialisation
71 The 'rom' file is special in that it provides read-only access to the device's
72 ROM file, if available. It's disabled by default, however, so applications
73 should write the string "1" to the file to enable it before attempting a read
74 call, and disable it following the access by writing "0" to the file. Note
75 that the device must be enabled for a rom read to return data successfully.
76 In the event a driver is not bound to the device, it can be enabled using the
77 'enable' file, documented above.
79 The 'remove' file is used to remove the PCI device, by writing a non-zero
80 integer to the file. This does not involve any kind of hot-plug functionality,
81 e.g. powering off the device. The device is removed from the kernel's list of
82 PCI devices, the sysfs directory for it is removed, and the device will be
83 removed from any drivers attached to it. Removal of PCI root buses is
86 Accessing legacy resources through sysfs
87 ----------------------------------------
89 Legacy I/O port and ISA memory resources are also provided in sysfs if the
90 underlying platform supports them. They're located in the PCI class hierarchy,
93 /sys/class/pci_bus/0000:17/
94 |-- bridge -> ../../../devices/pci0000:17
99 The legacy_io file is a read/write file that can be used by applications to
100 do legacy port I/O. The application should open the file, seek to the desired
101 port (e.g. 0x3e8) and do a read or a write of 1, 2 or 4 bytes. The legacy_mem
102 file should be mmapped with an offset corresponding to the memory offset
103 desired, e.g. 0xa0000 for the VGA frame buffer. The application can then
104 simply dereference the returned pointer (after checking for errors of course)
105 to access legacy memory space.
107 Supporting PCI access on new platforms
108 --------------------------------------
110 In order to support PCI resource mapping as described above, Linux platform
111 code must define HAVE_PCI_MMAP and provide a pci_mmap_page_range function.
112 Platforms are free to only support subsets of the mmap functionality, but
113 useful return codes should be provided.
115 Legacy resources are protected by the HAVE_PCI_LEGACY define. Platforms
116 wishing to support legacy functionality should define it and provide
117 pci_legacy_read, pci_legacy_write and pci_mmap_legacy_page_range functions.