1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
|
/**
*
* @file compat.h
* @author Michael Stapelberg
* @date 2009-04-05
* @brief Contains compatibility definitions for the different linux kernel versions to avoid
* putting ifdefs all over the driver code.
*
*/
#ifndef _COMPAT_H
#define _COMPAT_H
/* dev_name is the wrapper one needs to use to access what was formerly called
* bus_id in struct device. However, before 2.6.27, direct access was necessary,
* so we provide our own version. */
#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,27)
static inline const char *dev_name(struct device *dev) {
return dev->bus_id;
}
#endif
/* SetPageLocked disappeared in v2.6.27 */
#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,27)
#define compat_lock_page SetPageLocked
#define compat_unlock_page ClearPageLocked
#else
/* in v2.6.28, __set_page_locked and __clear_page_locked was introduced */
#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,28)
#define compat_lock_page __set_page_locked
#define compat_unlock_page __clear_page_locked
#else
/* However, in v2.6.27 itself, neither of them is there, so
* we need to use our own function fiddling with bits inside
* the page struct :-\ */
static inline void compat_lock_page(struct page *page) {
__set_bit(PG_locked, &page->flags);
}
static inline void compat_unlock_page(struct page *page) {
__clear_bit(PG_locked, &page->flags);
}
#endif
#endif
/* Before 2.6.13, simple_class was the standard interface. Nowadays, it's just called class */
#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,13)
#define class_compat class_simple
/* These functions are redirected to their old corresponding functions */
#define class_create(module, name) class_simple_create(module, name)
#define class_destroy(type) class_simple_destroy(type)
#define class_device_destroy(unused, devno) class_simple_device_remove(devno)
#define class_device_create(type, unused, devno, devpointer, nameformat, minor, unused) \
class_simple_device_add(type, devno, devpointer, nameformat, minor)
#define class_set_devdata(classdev, privdata) classdev->class_data = privdata
#define DEVICE_ATTR_COMPAT
#define sysfs_attr_def_name(name) class_device_attr_##name
#define sysfs_attr_def_pointer privdata->class_dev
#define SYSFS_GET_FUNCTION(name) ssize_t name(struct class_device *cls, char *buf)
#define SYSFS_SET_FUNCTION(name) ssize_t name(struct class_device *cls, const char *buf, size_t count)
#define SYSFS_GET_PRIVDATA (pcidriver_privdata_t*)cls->class_data
#else
/* In 2.6.26, device.h was changed quite significantly. Luckily, it only affected
type/function names, for the most part. */
//#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,26)
#define class_device_attribute device_attribute
#define CLASS_DEVICE_ATTR DEVICE_ATTR
#define class_device device
#define class_data dev
#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,27)
#define class_device_create(type, parent, devno, devpointer, nameformat, minor, privdata) \
device_create(type, parent, devno, privdata, nameformat, minor)
#else
#define class_device_create(type, parent, devno, devpointer, nameformat, minor, unused) \
device_create(type, parent, devno, nameformat, minor)
#endif
#define class_device_create_file device_create_file
#define class_device_remove_file device_remove_file
#define class_device_destroy device_destroy
#define DEVICE_ATTR_COMPAT struct device_attribute *attr,
#define class_set_devdata dev_set_drvdata
#define sysfs_attr_def_name(name) dev_attr_##name
#define sysfs_attr_def_pointer privdata->class_dev
#define SYSFS_GET_FUNCTION(name) ssize_t name(struct device *dev, struct device_attribute *attr, char *buf)
#define SYSFS_SET_FUNCTION(name) ssize_t name(struct device *dev, struct device_attribute *attr, const char *buf, size_t count)
#define SYSFS_GET_PRIVDATA dev_get_drvdata(dev)
//#endif
#define class_compat class
#endif
/* The arguments of IRQ handlers have been changed in 2.6.19. It's very likely that
int irq will disappear somewhen in the future (current is 2.6.29), too. */
#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,19)
#define IRQ_HANDLER_FUNC(name) irqreturn_t name(int irq, void *dev_id)
#else
#define IRQ_HANDLER_FUNC(name) irqreturn_t name(int irq, void *dev_id, struct pt_regs *regs)
#endif
/* atomic_inc_return appeared in 2.6.9, at least in CERN scientific linux, provide
compatibility wrapper for older kernels */
#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,9)
static int atomic_inc_return(atomic_t *variable) {
atomic_inc(variable);
return atomic_read(variable);
}
#endif
/* sg_set_page is available starting at 2.6.24 */
#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,24)
#define sg_set_page(sg, set_page, set_length, set_offset) do { \
(sg)->page = set_page; \
(sg)->length = set_length; \
(sg)->offset = set_offset; \
} while (0)
#endif
/* Before 2.6.20, disable was not an atomic counter, so this check was needed */
#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,20)
#define pci_disable_device(pdev) do { \
if (pdev->is_enabled) \
pci_disable_device(pdev); \
} while (0)
#endif
/* Before 2.6.24, scatter/gather lists did not need to be initialized */
#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,24)
#define sg_init_table(sg, nr_pages)
#endif
/* SA_SHIRQ was renamed to IRQF_SHARED in 2.6.24 */
#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,24)
#define request_irq(irq, irq_handler, modname, privdata) request_irq(irq, irq_handler, IRQF_SHARED, modname, privdata)
#else
#define request_irq(irq, irq_handler, modname, privdata) request_irq(irq, irq_handler, SA_SHIRQ, modname, privdata)
#endif
/* In 2.6.13, io_remap_page_range was removed in favor for io_remap_pfn_range which works on
more platforms and allows more memory space */
#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,13)
#define io_remap_pfn_range_compat(vmap, vm_start, bar_addr, bar_length, vm_page_prot) \
io_remap_pfn_range(vmap, vm_start, (bar_addr >> PAGE_SHIFT), bar_length, vm_page_prot)
#else
#define io_remap_pfn_range_compat(vmap, vm_start, bar_addr, bar_length, vm_page_prot) \
io_remap_page_range(vmap, vm_start, bar_addr, bar_length, vm_page_prot)
#endif
/* In 2.6.10, remap_pfn_range was introduced, see io_remap_pfn_range_compat */
#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,10)
#define remap_pfn_range_compat(vmap, vm_start, bar_addr, bar_length, vm_page_prot) \
remap_pfn_range(vmap, vm_start, (bar_addr >> PAGE_SHIFT), bar_length, vm_page_prot)
#define remap_pfn_range_cpua_compat(vmap, vm_start, cpua, size, vm_page_prot) \
remap_pfn_range(vmap, vm_start, page_to_pfn(virt_to_page((void*)cpua)), size, vm_page_prot)
#else
#define remap_pfn_range_compat(vmap, vm_start, bar_addr, bar_length, vm_page_prot) \
remap_page_range(vmap, vm_start, bar_addr, bar_length, vm_page_prot)
#define remap_pfn_range_cpua_compat(vmap, vm_start, cpua, size, vm_page_prot) \
remap_page_range(vmap, vm_start, virt_to_phys((void*)cpua), size, vm_page_prot)
#endif
/**
* Go over the pages of the kmem buffer, and mark them as reserved.
* This is needed, otherwise mmaping the kernel memory to user space
* will fail silently (mmaping /dev/null) when using remap_xx_range.
*/
static inline void set_pages_reserved_compat(unsigned long cpua, unsigned long size)
{
/* Starting in 2.6.15, the PG_RESERVED bit was removed.
See also http://lwn.net/Articles/161204/ */
#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,15)
struct page *page, *last_page;
page = virt_to_page(cpua);
last_page = virt_to_page(cpua + size - 1);
for (; page <= last_page; page++)
SetPageReserved(page);
#endif
}
#endif
|