Based on kernel version 4.16.1. Page generated on 2018-04-09 11:52 EST.
1 Console Drivers 2 =============== 3 4 The linux kernel has 2 general types of console drivers. The first type is 5 assigned by the kernel to all the virtual consoles during the boot process. 6 This type will be called 'system driver', and only one system driver is allowed 7 to exist. The system driver is persistent and it can never be unloaded, though 8 it may become inactive. 9 10 The second type has to be explicitly loaded and unloaded. This will be called 11 'modular driver' by this document. Multiple modular drivers can coexist at 12 any time with each driver sharing the console with other drivers including 13 the system driver. However, modular drivers cannot take over the console 14 that is currently occupied by another modular driver. (Exception: Drivers that 15 call do_take_over_console() will succeed in the takeover regardless of the type 16 of driver occupying the consoles.) They can only take over the console that is 17 occupied by the system driver. In the same token, if the modular driver is 18 released by the console, the system driver will take over. 19 20 Modular drivers, from the programmer's point of view, has to call: 21 22 do_take_over_console() - load and bind driver to console layer 23 give_up_console() - unload driver, it will only work if driver is fully unbond 24 25 In newer kernels, the following are also available: 26 27 do_register_con_driver() 28 do_unregister_con_driver() 29 30 If sysfs is enabled, the contents of /sys/class/vtconsole can be 31 examined. This shows the console backends currently registered by the 32 system which are named vtcon<n> where <n> is an integer from 0 to 15. Thus: 33 34 ls /sys/class/vtconsole 35 . .. vtcon0 vtcon1 36 37 Each directory in /sys/class/vtconsole has 3 files: 38 39 ls /sys/class/vtconsole/vtcon0 40 . .. bind name uevent 41 42 What do these files signify? 43 44 1. bind - this is a read/write file. It shows the status of the driver if 45 read, or acts to bind or unbind the driver to the virtual consoles 46 when written to. The possible values are: 47 48 0 - means the driver is not bound and if echo'ed, commands the driver 49 to unbind 50 51 1 - means the driver is bound and if echo'ed, commands the driver to 52 bind 53 54 2. name - read-only file. Shows the name of the driver in this format: 55 56 cat /sys/class/vtconsole/vtcon0/name 57 (S) VGA+ 58 59 '(S)' stands for a (S)ystem driver, ie, it cannot be directly 60 commanded to bind or unbind 61 62 'VGA+' is the name of the driver 63 64 cat /sys/class/vtconsole/vtcon1/name 65 (M) frame buffer device 66 67 In this case, '(M)' stands for a (M)odular driver, one that can be 68 directly commanded to bind or unbind. 69 70 3. uevent - ignore this file 71 72 When unbinding, the modular driver is detached first, and then the system 73 driver takes over the consoles vacated by the driver. Binding, on the other 74 hand, will bind the driver to the consoles that are currently occupied by a 75 system driver. 76 77 NOTE1: Binding and unbinding must be selected in Kconfig. It's under: 78 79 Device Drivers -> Character devices -> Support for binding and unbinding 80 console drivers 81 82 NOTE2: If any of the virtual consoles are in KD_GRAPHICS mode, then binding or 83 unbinding will not succeed. An example of an application that sets the console 84 to KD_GRAPHICS is X. 85 86 How useful is this feature? This is very useful for console driver 87 developers. By unbinding the driver from the console layer, one can unload the 88 driver, make changes, recompile, reload and rebind the driver without any need 89 for rebooting the kernel. For regular users who may want to switch from 90 framebuffer console to VGA console and vice versa, this feature also makes 91 this possible. (NOTE NOTE NOTE: Please read fbcon.txt under Documentation/fb 92 for more details). 93 94 Notes for developers: 95 ===================== 96 97 do_take_over_console() is now broken up into: 98 99 do_register_con_driver() 100 do_bind_con_driver() - private function 101 102 give_up_console() is a wrapper to do_unregister_con_driver(), and a driver must 103 be fully unbound for this call to succeed. con_is_bound() will check if the 104 driver is bound or not. 105 106 Guidelines for console driver writers: 107 ===================================== 108 109 In order for binding to and unbinding from the console to properly work, 110 console drivers must follow these guidelines: 111 112 1. All drivers, except system drivers, must call either do_register_con_driver() 113 or do_take_over_console(). do_register_con_driver() will just add the driver to 114 the console's internal list. It won't take over the 115 console. do_take_over_console(), as it name implies, will also take over (or 116 bind to) the console. 117 118 2. All resources allocated during con->con_init() must be released in 119 con->con_deinit(). 120 121 3. All resources allocated in con->con_startup() must be released when the 122 driver, which was previously bound, becomes unbound. The console layer 123 does not have a complementary call to con->con_startup() so it's up to the 124 driver to check when it's legal to release these resources. Calling 125 con_is_bound() in con->con_deinit() will help. If the call returned 126 false(), then it's safe to release the resources. This balance has to be 127 ensured because con->con_startup() can be called again when a request to 128 rebind the driver to the console arrives. 129 130 4. Upon exit of the driver, ensure that the driver is totally unbound. If the 131 condition is satisfied, then the driver must call do_unregister_con_driver() 132 or give_up_console(). 133 134 5. do_unregister_con_driver() can also be called on conditions which make it 135 impossible for the driver to service console requests. This can happen 136 with the framebuffer console that suddenly lost all of its drivers. 137 138 The current crop of console drivers should still work correctly, but binding 139 and unbinding them may cause problems. With minimal fixes, these drivers can 140 be made to work correctly. 141 142 ========================== 143 Antonino Daplas <adaplas@pol.net>