A derogatory term for an insurgent fighter who trended to be black. Originated with Rhodesian military personnel in Rhodesia (currently Zimbabwe) during the Rhodesian Bush War. The term arises from how insurgents would flop down dead when shot. A derogatory term for an insurgent fighter who trended to be black. Originated with Rhodesian military personnel in Rhodesia (currently Zimbabwe) during the Rhodesian Bush War. The term arises from how insurgents would flop down dead when shot.

• Slot Floppies Definition History
• Slot Floppies Definition Deep

(redirected from floppies)
Also found in: Thesaurus, Financial, Idioms, Encyclopedia.

flop·py

(flŏp′ē)adj.flop·pi·er, flop·pi·est

floppy

(ˈflɒpɪ)

flop•py

(ˈflɒp i)
adj. -pi•er, -pi•est,adj. n.

Noun	1.	floppy - a small plastic magnetic disk enclosed in a stiff envelope with a radial slit; used to store data or programs for a microcomputer; 'floppy disks are noted for their relatively slow speed and small capacity and low price' diskette, floppy disk computer, computing device, computing machine, data processor, electronic computer, information processing system - a machine for performing calculations automatically magnetic disc, magnetic disk, disc, disk - (computer science) a memory device consisting of a flat disk covered with a magnetic coating on which information is stored
Adj.	1.	floppy - hanging limply; 'a spaniel with floppy ears' lax - lacking in strength or firmness or resilience; 'a lax rope'; 'a limp handshake'

floppy

floppy

adjective

floppy

[ˈflɒpɪ]

floppy

[ˈflɒpi]adj [hair, fringe] → lâche, flottant(e)
a floppy hat → un chapeau à bordsflottantsfloppy disk n(COMPUTING) → disquettef

floppy

floppy

[ˈflɒpɪ]adj (-ier (comp) (-iest (superl))) → floscio/a, molle

Slot Floppies Definition History

floppy hat → cappello floscio

flop

(flop) verb – past tense, past participle flopped – nounˈfloppy adjectiveˌfloppy ˈdisk noun
Want to thank TFD for its existence? Tell a friend about us, add a link to this page, or visit the webmaster's page for free fun content.

Slot Floppies Definition Deep

Link to this page:
TheHyperNews Linux KHGDiscussion Pages

Device Driver Basics

We will assume that you decide that you do not wish to writea user-space device, and would rather implement your device inthe kernel. You will probably be writing writing two files, a.c file and a .h file, and possibly modifyingother files as well, as will be described below. We will referto your files as foo.c and foo.h, and your driver will be thefoo driver.

Namespace

One of the first things you will need to do, before writingany code, is to name your device. This name should be a short(probably two or three character) string. For instance, theparallel device is the ``lp' device, the floppies arethe ``fd' devices, and SCSI disks are the``sd' devices. As you write your driver, you willgive your functions names prefixed with your chosen string toavoid any namespace confusion. We will call your prefixfoo, and give your functions names likefoo_read(), foo_write(), etc.

Allocating memory

Memory allocation in the kernel is a little different frommemory allocation in normal user-level programs. Instead ofhaving a malloc() capable of delivering almostunlimited amounts of memory, there is a kmalloc()function that is a bit different:

• Memory is provided in pieces whose size is a power of 2, exceptthat pieces larger than 128 bytes are allocated in blocks whose sizeis a power of 2 minus some small amount for overhead.You can request any odd size, but memory will not be used any moreefficiently if you request a 31-byte piece than it will if you requesta 32 byte piece. Also, there is a limit to the amount of memory thatcan be allocated, which is currently 131056 bytes.
• kmalloc() takes a second argument, the priority. This isused as an argument to the get_free_page() function, where itis used to determine when to return. The usual priority isGFP_KERNEL. If it may be called from within an interrupt, useGFP_ATOMIC and be truly prepared for it to fail (don't panic).This is because if you specify GFP_KERNEL, kmalloc() maysleep, which cannot be done on an interrupt. The other option isGFP_BUFFER, which is used only when the kernel is allocating bufferspace, and never in device drivers.

To free memory allocated with kmalloc(), use one oftwo functions: kfree() or kfree_s(). Thesediffer from free() in a few ways as well:

• kfree() is a macro which calls kfree_s() and actslike the standard free() outside the kernel.
• If you know what size object you are freeing, you can speedthings up by calling kfree_s() directly. It takes twoarguments: the first is the pointer that you are freeing, as in thesingle argument to kfree(), and the second is the size of theobject being freed.

See Supporting Functions formore information on kmalloc(), kfree(), andother useful functions.

Be gentle when you use kmalloc. Use only what you have to.Remember that kernel memory is unswappable, and thus allocatingextra memory in the kernel is a far worse thing to do in thekernel than in a user-level program. Take only what you need,and free it when you are done, unless you are going to use itright away again.

Character vs. block devices

There are two main types of devices under all Unix systems,character and block devices. Character devices are those forwhich no buffering is performed, and block devices are thosewhich are accessed through a cache. Block devices must berandom access, but character devices are not required to be,though some are. Filesystems can only be mounted if they areon block devices.

Character devices are read from and written to with twofunction: foo_read() and foo_write(). Theread() and write() calls do not return untilthe operation is complete. By contrast, block devices do noteven implement the read() and write()functions, and instead have a function which has historicallybeen called the ``strategy routine.' Reads and writes are donethrough the buffer cache mechanism by the generic functionsbread(),breada(), and bwrite().These functions go through the buffer cache, and so may or maynot actually call the strategy routine, depending on whether ornot the block requested is in the buffer cache (for reads) oron whether or not the buffer cache is full (for writes). Arequest may be asyncronous: breada() can request thestrategy routine to schedule reads that have not been askedfor, and to do it asyncronously, in the background, in thehopes that they will be needed later.

The sources for character devices are kept in drivers/char/,and the sources for block devices are kept in drivers/block/.They have similar interfaces, and are very much alike, exceptfor reading and writing. Because of the difference in readingand writing, initialization is different, as block devices haveto register a strategy routine, which is registered in adifferent way than the foo_read() andfoo_write() routines of a character device driver.Specifics are dealt with inCharacter Device Initialization andBlock Device Initialization.

Interrupts vs. Polling

Hardware is slow. That is, in the time it takes to getinformation from your average device, the CPU could be offdoing something far more useful than waiting for a busy butslow device. So to keep from having to busy-wait allthe time, interrupts are provided which can interruptwhatever is happening so that the operating system can do sometask and return to what it was doing without losinginformation. In an ideal world, all devices would probablywork by using interrupts. However, on a PC or clone, there areonly a few interrupts available for use by your peripherals, sosome drivers have to poll the hardware: ask the hardware if itis ready to transfer data yet. This unfortunately wastes time,but it sometimes needs to be done.

Some hardware (like memory-mapped displays) is as fastas the rest of the machine, and does not generate outputasyncronously, so an interrupt-driven driver would be rathersilly, even if interrupts were provided.

In Linux, many of the drivers are interrupt-driven, but someare not, and at least one can be either, and can be switchedback and forth at runtime. For instance, the lpdevice (the parallel port driver) normally polls the printer tosee if the printer is ready to accept output, and if theprinter stays in a not ready phase for too long, the driverwill sleep for a while, and try again later. This improvessystem performance. However, if you have a parallel card thatsupplies an interrupt, the driver will utilize that, which willusually make performance even better.

There are some important programming differences betweeninterrupt-driven drivers and polling drivers. To understandthis difference, you have to understand a little bit of howsystem calls work under Unix. The kernel is not a separatetask under Unix. Rather, it is as if each process has a copy ofthe kernel. When a process executes a system call, it does nottransfer control to another process, but rather, the processchanges execution modes, and is said to be ``in kernel mode.'In this mode, it executes kernel code which is trusted to besafe.

In kernel mode, the process can still access the user-spacememory that it was previously executing in, which is donethrough a set of macros: get_fs_*() andmemcpy_fromfs() read user-space memory, andput_fs_*() and memcpy_tofs() write touser-space memory. Because the process is still running, butin a different mode, there is no question of where in memory toput the data, or where to get it from. However, when aninterrupt occurs, any process might currently be running, sothese macros cannot be used--if they are, they will eitherwrite over random memory space of the running process or causethe kernel to panic.

Instead, when scheduling the interrupt, a driver must alsoprovide temporary space in which to put the information, andthen sleep. When the interrupt-driven part of the driver hasfilled up that temporary space, it wakes up the process, whichcopies the information from that temporary space into theprocess' user space and returns. In a block device driver,this temporary space is automatically provided by the buffercache mechanism, but in a character device driver, the driveris responsible for allocating it itself.

The sleep-wakeup mechanism

[Begin by giving a general description of how sleeping is usedand what it does. This should mention things like all processessleeping on an event are woken at once, and then they contend for theevent again, etc...]

Perhaps the best way to try to understand the Linuxsleep-wakeup mechanism is to read the source for the__sleep_on() function, used to implement both thesleep_on() and interruptible_sleep_on()calls.

A wait_queue is a circular list of pointers to taskstructures, defined in <linux/wait.h> to bestate is either TASK_INTERRUPTIBLE orTASK_UNINTERUPTIBLE, depending on whether or not thesleep should be interruptable by such things as system calls.In general, the sleep should be interruptible if the device isa slow one; one which can block indefinitely, includingterminals and network devices or pseudodevices.

add_wait_queue() turns off interrupts, if they wereenabled, and adds the new struct wait_queue declaredat the beginning of the function to the list p. Itthen recovers the original interrupt state (enabled ordisabled), and returns.

save_flags() is a macro which saves the processflags in its argument. This is done to preserve the previousstate of the interrupt enable flag. This way, therestore_flags() later can restore the interrupt state,whether it was enabled or disabled. sti() then allowsinterrupts to occur, and schedule() finds a newprocess to run, and switches to it. Schedule will not choosethis process to run again until the state is changed toTASK_RUNNING by wake_up() called on the samewait queue, p, or conceivably by something else.

The process then removes itself from thewait_queue, restores the orginal interrupt conditionwith restore_flags(), and returns.

Whenever contention for a resource might occur, there needsto be a pointer to a wait_queue associated with thatresource. Then, whenever contention does occur, each processthat finds itself locked out of access to the resource sleepson that resource's wait_queue. When any process isfinished using a resource for which there is await_queue, it should wake up and processes that mightbe sleeping on that wait_queue, probably by callingwake_up(), or possiblywake_up_interruptible().

If you don't understand why a process might want to sleep,or want more details on when and how to structure thissleeping, I urge you to buy one of the operating systemstextbooks listed in theAnnotated Bibliography and look upmutual exclusion and deadlock.

More advanced sleeping

If the sleep_on()/wake_up() mechanism inLinux does not satisfy your device driver needs, you can codeyour own versions of sleep_on() and wake_up()that fit your needs. For an example of this, look at theserial device driver (drivers/char/serial.c) in functionblock_til_ready(), where quite a bit has to be donebetween the add_wait_queue() and theschedule().

The VFS

The Virtual Filesystem Switch, or VFS, is themechanism which allows Linux to mount many differentfilesystems at the same time. In the first versions of Linux,all filesystem access went straight into routines whichunderstood the minix filesystem. To make it possiblefor other filesystems to be written, filesystem calls had topass through a layer of indirection which would switch the callto the routine for the correct filesystem. This was done bysome generic code which can handle generic cases and astructure of pointers to functions which handle specific cases.One structure is of interest to the device driver writer; thefile_operations structure.

From /usr/include/linux/fs.h:Essentially, this structure constitutes a parital list ofthe functions that you may have to write to create your driver.

This section details the actions and requirements of thefunctions in the file_operations structure. Itdocuments all the arguments that these functions take. [Itshould also detail all the defaults, and cover more carefullythe possible return values.]

The lseek() function

This function is called when the system calllseek() is called on the device special filerepresenting your device. An understanding of what the systemcall lseek() does should be sufficient to explain thisfunction, which moves to the desired offset. It takes thesefour arguments:

struct inode * inode

Pointer to the inode structure for this device.

struct file * file

Pointer to the file structure for this device.

off_t offset

Offset from origin to move to.

int origin

0 = take the offset from absolute offset 0 (the beginning).
1 = take the offset from the current position.
2 = take the offset from the end.

lseek() returns -errno on error, or the absoluteposition (>= 0) after the lseek.

If there is no lseek(), the kernel will take thedefault action, which is to modify the file->f_poselement. For an origin of 2, the default action is toreturn -EINVAL if file->f_inode is NULL,otherwise it sets file->f_pos tofile->f_inode->i_size + offset. Because ofthis, if lseek() should return an error for yourdevice, you must write an lseek() function whichreturns that error.

The read() and write() functions

The read and write functions read and write a characterstring to the device. If there is no read() orwrite() function in the file_operationsstructure registered with the kernel, and the device is acharacter device, read() or write() systemcalls, respectively, will return -EINVAL. If thedevice is a block device, these functions should not beimplemented, as the VFS will route requests through the buffercache, which will call your strategy routine. Theread and write functions take thesearguments:

struct inode * inode

This is a pointer to the inode of the device special file which wasaccessed. From this, you can do several things, based on thestruct inode declaration about 100 lines into/usr/include/linux/fs.h. For instance, you can find the minor numberof the file by this construction: unsigned int minor =MINOR(inode->i_rdev); The definition of the MINOR macro is in, as are many other useful definitions. Read fs.hand a few device drivers for more details, and seeSupporting Functionsfor a short description. inode->i_modecan be used to find the mode of the file, and there are macrosavailable for this, as well.

struct file * file

Pointer to file structure for this device.

char * buf

This is a buffer of characters to read or write. It islocated in user-space memory, and therefore must beaccessed using the get_fs*(), put_fs*(), andmemcpy*fs() macros detailed in Supporting Functions. User-spacememory is inaccessible during an interrupt, so if your driveris interrupt driven, you will have to copy the contents of yourbuffer into a queue.

int count

This is a count of characters in buf tobe read or written. It is the size of buf, and is how you knowthat you have reached the end of buf, as buf is notguaranteed to be null-terminated.

The readdir() function

This function is another artifact offile_operations being used for implementingfilesystems as well as device drivers. Do not implement it.The kernel will return -ENOTDIR if the system callreaddir() is called on your device special file.

The select() function

The select() function is generally most useful withcharacter devices. It is usually used to multiplex readswithout polling--the application calls the select()system call, giving it a list of file descriptors to watch, andthe kernel reports back to the program on which file descriptorhas woken it up. It is also used as a timer. However, theselect() function in your device driver is notdirectly called by the system call select(), and sothe file_operationsselect() only needs to doa few things. Its arguments are:

struct inode * inode

Pointer to the inode structure for this device.

struct file * file

Pointer to the file structure for this device.

int sel_type

The select type to perform:sys_select(), which originall called yourselect() function, uses the information given to it bythe select_wait() function to put the process tosleep. select_wait() adds the process to the waitqueue, but do_select() (called fromsys_select()) actually puts the process to sleep bychanging the process state to TASK_INTERRUPTIBLE andcalling schedule().

The first argument to select_wait() is the samewait_queue that should be used for asleep_on(), and the second is theselect_table that was passed to your select()function.

After having explained all this in excruciating detail, hereare two rules to follow:

1. Call select_wait() if the device is not ready, and return 0.
2. Return 1 if the device is ready.

If you provide a select() function, do not providetimeouts by setting current->timeout, as theselect() mechanism uses current->timeout,and the two methods cannot co-exist, as there is only onetimeout for each process. Instead, consider using atimer to provide timeouts. See the description of theadd_timer() function inSupporting Functionsfor details.

The ioctl() function

The ioctl() function processes ioctl calls. Thestructure of your ioctl() function will be: firsterror checking, then one giant (possibly nested) switchstatement to handle all possible ioctls. The ioctl number ispassed as cmd, and the argument to the ioctl is passedas arg. It is good to have an understanding of howioctls ought to work before making them up. If youare not sure about your ioctls, do not feel ashamed to asksomeone knowledgeable about it, for a few reasons: you may noteven need an ioctl for your purpose, and if you do need anioctl, there may be a better way to do it than what you havethought of. Since ioctls are the least regular part of thedevice interface, it takes perhaps the most work to get thispart right. Take the time and energy you need to get it right.

The first thing you need to do is look inDocumentation/ioctl-number.txt, read it, and pick an unused number.Then go from there.

struct inode * inode

Pointer to the inode structure for this device.

struct file * file

Pointer to the file structure for this device.

unsigned int cmd

This is the ioctl command. It isgenerally used as the switch variable for a case statement.

unsigned int arg

This is the argument to the command.This is user defined. Since this is the same size as a(void *), this can be used as a pointer to user space, accessedthrough the fs register as usual.

Returns:

-errno on error
Every other return is user-defined.

If the ioctl() slot in the file_operations structure isnot filled in, the VFS will return -EINVAL. However, inall cases, if cmd is one of FIOCLEX, FIONCLEX,FIONBIO, or FIOASYNC, default processing will be done:

FIOCLEX (0x5451)

Sets the close-on-exec bit.

FIONCLEX (0x5450)

Clears the close-on-exec bit.

FIONBIO (0x5421)

If arg is non-zero, set O_NONBLOCK,otherwise clear O_NONBLOCK.

FIOASYNC (0x5452)

If arg is non-zero, set O_SYNC,otherwise clear O_SYNC. O_SYNC is not yet implemented,but it is documented here and parsed in the kernel for completeness.

Note that you have to avoid these four numbers when creatingyour own ioctls, since if they conflict, the VFS ioctl codewill interpret them as being one of these four, and actappropriately, causing a very hard-to-track-down bug.

The mmap() function

struct inode * inode

Pointer to inode structure for device.

struct file * file

Pointer to file structure for device.

unsigned long addr

Beginning of address in main memoryto mmap() into.

size_t len

Length of memory to mmap().

int prot

One of:release()function is defined, none is called.

The init() function

This function is not actually included in thefile_operations structure, but you are required toimplement it, because it is this function that registers thefile_operations structure with the VFS in the firstplace--without this function, the VFS could not route anyrequests to the driver. This function is called when thekernel first boots and is configuring itself. The initfunction then detects all devices. You will have to call yourinit() function from the correct place: for acharacter device, this is chr_dev_init() indrivers/char/mem.c.

While the init() function runs, it registers yourdriver by calling the proper registration function. Forcharacter devices, this is register_chrdev(). (SeeSupporting Functionsfor more information on the registration functions.)register_chrdev() takes three arguments: the majordevice number (an int), the ``name' of the device (a string),and the address of the device_fopsfile_operations structure.

When this is done, and a character or block special file isaccessed, the VFS filesystem switch automagically routes thecall, whatever it is, to the proper function, if a functionexists. If the function does not exist, the VFS routines takesome default action.

The init() function usually displays someinformation about the driver, and usually reports all hardwarefound. All reporting is done via the printk()function.

Copyright (C) 1992, 1993, 1994, 1996 Michael K. Johnson,johnsonm@redhat.com.

Messages