projects / apple / xnu.git / blame
| Commit | Line | Data |
|---|---|---|
| 0b4e3aa0 A |
1 | 12345678901234567890123456789012345678901234567890123456789012345678901234567890 |
| 2 | ||
| 3 | Description of User Routines in Prngcore | |
| 4 | ---------------------------------------------- | |
| 5 | ||
| 6 | This files describes routines in prngcore that are designed to be called by the | |
| 7 | user (ie client apps). Those interested in the details of the library are | |
| 8 | directed to readme-prngcoder. | |
| 9 | ||
| 10 | Files of interest in this directory | |
| 11 | ----------------------------------- | |
| 12 | ||
| 13 | yarrow.h | |
| 14 | ||
| 15 | Main header file (and the only one needed) for client apps. | |
| 16 | ||
| 17 | userdefines.h | |
| 18 | ||
| 19 | Header file with macros that can be defined to specify the system that this | |
| 20 | code is being compiled on, as well as other details of the prng operation. | |
| 21 | ||
| 22 | usersources.h | |
| 23 | ||
| 24 | Header file containing the names of the various user sources of entropic data. | |
| 25 | You can add/delete/rename sources by altering the entries in the enumeration. | |
| 26 | ||
| 27 | ||
| 28 | PRNG Client Routines | |
| 29 | -------------------- | |
| 30 | ||
| 31 | All major routines return the success/error value for their operation. | |
| 32 | ||
| 33 | prngOutput(outbuf,outbuflen) | |
| 34 | ||
| 35 | Writes outbuflen worth of "random" data to outbuf. This routine has | |
| 36 | backtracking protection, but you should call prngAllowReseed whenever you can | |
| 37 | spare the cycles to guarantee good output. | |
| 38 | ||
| 39 | prngStretch(inbuf,inbuflen,outbuf,outbuflen) | |
| 40 | ||
| 41 | Takes inbuflen bytes of data from inbuf and turns it into outbuflen bytes of | |
| 42 | data stored in outbuf. | |
| 43 | ||
| 44 | prngInput(inbuf,inbuflen,poolnum,estbits) | |
| 45 | ||
| 46 | Takes inbuflen bytes of data from inbuf and places it in entropy pool poolnum. | |
| 47 | The user entropy pool names can be found in usersources.h (see above). | |
| 48 | ||
| 49 | prngForceReseed(ticks) | |
| 50 | ||
| 51 | Forces a reseed that lasts about ticks ticks long. Be very careful when using | |
| 52 | this function to ensure that you do not produce a poor output state. It is | |
| 53 | suggested that you instead use prngAllowReseed. | |
| 54 | ||
| 55 | prngAllowReseed(ticks) | |
| 56 | ||
| 57 | Will force a reseed if there is enough entropy. A reseed (of length ticks) | |
| 58 | will be done if the total entropy estimate, ignoring the K greatest sources, | |
| 59 | is greater than THRESHOLD. Currently, K = 0 (a bad idea) and THRESHOLD = 100 | |
| 60 | (likely to remain so). These values can be found and edited in userdefines.h. | |
| 61 | Will return PRNG_ERR_NOT_ENOUGH_ENTROPY if there is not enough entropy in the | |
| 62 | pool at this time. | |
| 63 | ||
| 64 | prngProcessSeedBuffer(buf,ticks) | |
| 65 | ||
| 66 | Takes 20 bytes of data from buf and churns it into the entropy pool, and then | |
| 67 | forces a reseed of length ticks. The first 20 bytes of output are then | |
| 68 | returned in buf for future use with this function. It is recommended that data | |
| 69 | used with this function be stored very securely. | |
| 70 | ||
| 71 | prngSlowPoll(pollsize) | |
| 72 | ||
| 73 | Does a slow poll to collect a large amount of vaguely random data from the OS | |
| 74 | itself. The poll with collect at most pollsize bytes, and this parameter can | |
| 75 | be used to control (approximately) the length of the poll. The collected data | |
| 76 | is fed into the entropy pool. After calling this function you may call either | |
| 77 | allow (recommended) or force a reseed if desired. | |
| 78 | ||
| 79 | -------- | |
| 80 | ||
| 81 | Any questions can be directed to the programmer (me), Ari Benbasat, at | |
| 82 | pigsfly@unixg.ubc.ca. Comments would be greatly appreciated. Please cc: all | |
| 83 | e-mail to Bruce Schneier, John Kelsey and Chris Hall | |
| 84 | {schneier,kelsey,hall}@counterpane.com. | |
| 85 | ||
| 86 | Thank you. | |
| 87 | ||
| 88 | ||
| 89 | ||
| 90 | i |