1 =========== 2 ISA Drivers 3 =========== 4 5 The following text is adapted from the commit message of the initial 6 commit of the ISA bus driver authored by Rene Herman. 7 8 During the recent "isa drivers using platform devices" discussion it was 9 pointed out that (ALSA) ISA drivers ran into the problem of not having 10 the option to fail driver load (device registration rather) upon not 11 finding their hardware due to a probe() error not being passed up 12 through the driver model. In the course of that, I suggested a separate 13 ISA bus might be best; Russell King agreed and suggested this bus could 14 use the .match() method for the actual device discovery. 15 16 The attached does this. For this old non (generically) discoverable ISA 17 hardware only the driver itself can do discovery so as a difference with 18 the platform_bus, this isa_bus also distributes match() up to the 19 driver. 20 21 As another difference: these devices only exist in the driver model due 22 to the driver creating them because it might want to drive them, meaning 23 that all device creation has been made internal as well. 24 25 The usage model this provides is nice, and has been acked from the ALSA 26 side by Takashi Iwai and Jaroslav Kysela. The ALSA driver module_init's 27 now (for oldisa-only drivers) become:: 28 29 static int __init alsa_card_foo_init(void) 30 { 31 return isa_register_driver(&snd_foo_isa_driver, SNDRV_CARDS); 32 } 33 34 static void __exit alsa_card_foo_exit(void) 35 { 36 isa_unregister_driver(&snd_foo_isa_driver); 37 } 38 39 Quite like the other bus models therefore. This removes a lot of 40 duplicated init code from the ALSA ISA drivers. 41 42 The passed in isa_driver struct is the regular driver struct embedding a 43 struct device_driver, the normal probe/remove/shutdown/suspend/resume 44 callbacks, and as indicated that .match callback. 45 46 The "SNDRV_CARDS" you see being passed in is a "unsigned int ndev" 47 parameter, indicating how many devices to create and call our methods 48 with. 49 50 The platform_driver callbacks are called with a platform_device param; 51 the isa_driver callbacks are being called with a ``struct device *dev, 52 unsigned int id`` pair directly -- with the device creation completely 53 internal to the bus it's much cleaner to not leak isa_dev's by passing 54 them in at all. The id is the only thing we ever want other then the 55 struct device anyways, and it makes for nicer code in the callbacks as 56 well. 57 58 With this additional .match() callback ISA drivers have all options. If 59 ALSA would want to keep the old non-load behaviour, it could stick all 60 of the old .probe in .match, which would only keep them registered after 61 everything was found to be present and accounted for. If it wanted the 62 behaviour of always loading as it inadvertently did for a bit after the 63 changeover to platform devices, it could just not provide a .match() and 64 do everything in .probe() as before. 65 66 If it, as Takashi Iwai already suggested earlier as a way of following 67 the model from saner buses more closely, wants to load when a later bind 68 could conceivably succeed, it could use .match() for the prerequisites 69 (such as checking the user wants the card enabled and that port/irq/dma 70 values have been passed in) and .probe() for everything else. This is 71 the nicest model. 72 73 To the code... 74 75 This exports only two functions; isa_{,un}register_driver(). 76 77 isa_register_driver() register's the struct device_driver, and then 78 loops over the passed in ndev creating devices and registering them. 79 This causes the bus match method to be called for them, which is:: 80 81 int isa_bus_match(struct device *dev, struct device_driver *driver) 82 { 83 struct isa_driver *isa_driver = to_isa_driver(driver); 84 85 if (dev->platform_data == isa_driver) { 86 if (!isa_driver->match || 87 isa_driver->match(dev, to_isa_dev(dev)->id)) 88 return 1; 89 dev->platform_data = NULL; 90 } 91 return 0; 92 } 93 94 The first thing this does is check if this device is in fact one of this 95 driver's devices by seeing if the device's platform_data pointer is set 96 to this driver. Platform devices compare strings, but we don't need to 97 do that with everything being internal, so isa_register_driver() abuses 98 dev->platform_data as a isa_driver pointer which we can then check here. 99 I believe platform_data is available for this, but if rather not, moving 100 the isa_driver pointer to the private struct isa_dev is ofcourse fine as 101 well. 102 103 Then, if the driver did not provide a .match, it matches. If it did, 104 the driver match() method is called to determine a match. 105 106 If it did **not** match, dev->platform_data is reset to indicate this to 107 isa_register_driver which can then unregister the device again. 108 109 If during all this, there's any error, or no devices matched at all 110 everything is backed out again and the error, or -ENODEV, is returned. 111 112 isa_unregister_driver() just unregisters the matched devices and the 113 driver itself. 114 115 module_isa_driver is a helper macro for ISA drivers which do not do 116 anything special in module init/exit. This eliminates a lot of 117 boilerplate code. Each module may only use this macro once, and calling 118 it replaces module_init and module_exit. 119 120 max_num_isa_dev is a macro to determine the maximum possible number of 121 ISA devices which may be registered in the I/O port address space given 122 the address extent of the ISA devices.
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.