FreeCalypso > hg > fc-selenite
view src/libsys/rand.c @ 112:fdecfb3bd860
.../drv_app/r2d/r2d_*.c: LCD include case fixes from Magnetite
R2D is not used in Selenite, but we strive to keep the two source trees
in sync as much as possible to reduce the diffs.
| author | Mychaela Falconia <falcon@freecalypso.org> |
|---|---|
| date | Sat, 13 Oct 2018 16:56:22 +0000 |
| parents | 425ab6d987f3 |
| children |
line wrap: on
line source
/* * This version of rand() has been lifted from Ancient UNIX, and modified * to match the version used in TI's TCS211 GSM firmware, as revealed by * disassembly of rand.obj in the rts16le_flash.lib binary library used * by that semi-src package. TI's version (most likely from their compiler * tools group, rather than the GSM fw group, but who knows) uses the * same trivial implementation of rand() as the original Ancient UNIX libc, * but with one change: TI's return value is right-shifted by 16 bits * compared to what the Ancient UNIX rand() would have returned. * The caller thus gets back only 15 pseudo-random bits rather than 31, * but then the lower bits of the original rand() return value are * known to be pretty bad. * * rand() is used by some FFS code in reclaim.c. If we don't provide our * own version of rand() and let the linker pull the version from newlib, * the link fails because the latter uses malloc. This ancient implementation * of rand() is quite poor, but my plan is to look into possibly adopting * some better PRNG after we get the basic TI GSM firmware reintegrated. */ static long randx = 1; srand(x) unsigned x; { randx = x; } rand() { return ((randx = randx * 1103515245 + 12345) & 0x7fffffff) >> 16; }
