d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Add a truely evil workaround (hack!) for some unfortunate BIOS

Browse Source 
programming practices.  It seems that newer fdc chips have an
alternative way of setting the transfer speed (including high speed
modes for floppy tape) that doesn't use the control register (which
we don't support - we use the old way only).  So, they (the BIOS
programmers) sometimes leave out the 0x3f6 control register from
the PnP ports descriptor(!!).  "Hey, it works with windows, so
what's the problem?" :-(  Anyway, this hack tries to compensate
for that.  This was discussed with dfr (who did the pnp attachment).
This commit is contained in:
peter 1999-12-06 06:59:09 +00:00
parent 8bc1a5a150
commit f57de250e0
2 changed files with 46 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

29
sys/dev/fdc/fdc.c
View File

@ -640,13 +640,30 @@ fdc_probe(device_t dev)
fdc->portt = rman_get_bustag(fdc->res_ioport);
fdc->porth = rman_get_bushandle(fdc->res_ioport);
/*
* Some bios' report the device at 0x3f2-0x3f5,0x3f7 and some at
* 0x3f0-0x3f5,0x3f7. We detect the former by checking the size
* and adjust the port address accordingly.
*
* And some (!!) report 0x3f2-0x3f5 and completely leave out the
* control register! It seems that some non-antique controller chips
* have a different method of programming the transfer speed which
* doesn't require the control register, but it's mighty bogus as the
* chip still responds to the address for the control register.
* This hack is truely evil as we use the 6th port in a 4-port chunk.
*/
/* 0x3f2-0x3f5 */
if (bus_get_resource_count(dev, SYS_RES_IOPORT, 0) == 4 &&
bus_get_resource_count(dev, SYS_RES_IOPORT, 1) == 0) {
fdc->port_off = -2;
ispnp = 0; /* hack, don't reserve second port chunk */
}
/* 0x3f0-0x3f5 */
if (bus_get_resource_count(dev, SYS_RES_IOPORT, 0) == 6 &&
bus_get_resource_count(dev, SYS_RES_IOPORT, 1) == 0) {
ispnp = 0; /* hack, don't reserve second port chunk */
}
if (ispnp) {
/*
* Some bios' report the device at 0x3f2-0x3f5,0x3f7
* and some at 0x3f0-0x3f5,0x3f7. We detect the former
* by checking the size and adjust the port address
* accordingly.
*/
if (bus_get_resource_count(dev, SYS_RES_IOPORT, 0) == 4)
fdc->port_off = -2;
fdc->flags |= FDC_ISPNP;

29
sys/isa/fd.c
View File

@ -640,13 +640,30 @@ fdc_probe(device_t dev)
fdc->portt = rman_get_bustag(fdc->res_ioport);
fdc->porth = rman_get_bushandle(fdc->res_ioport);
/*
* Some bios' report the device at 0x3f2-0x3f5,0x3f7 and some at
* 0x3f0-0x3f5,0x3f7. We detect the former by checking the size
* and adjust the port address accordingly.
*
* And some (!!) report 0x3f2-0x3f5 and completely leave out the
* control register! It seems that some non-antique controller chips
* have a different method of programming the transfer speed which
* doesn't require the control register, but it's mighty bogus as the
* chip still responds to the address for the control register.
* This hack is truely evil as we use the 6th port in a 4-port chunk.
*/
/* 0x3f2-0x3f5 */
if (bus_get_resource_count(dev, SYS_RES_IOPORT, 0) == 4 &&
bus_get_resource_count(dev, SYS_RES_IOPORT, 1) == 0) {
fdc->port_off = -2;
ispnp = 0; /* hack, don't reserve second port chunk */
}
/* 0x3f0-0x3f5 */
if (bus_get_resource_count(dev, SYS_RES_IOPORT, 0) == 6 &&
bus_get_resource_count(dev, SYS_RES_IOPORT, 1) == 0) {
ispnp = 0; /* hack, don't reserve second port chunk */
}
if (ispnp) {
/*
* Some bios' report the device at 0x3f2-0x3f5,0x3f7
* and some at 0x3f0-0x3f5,0x3f7. We detect the former
* by checking the size and adjust the port address
* accordingly.
*/
if (bus_get_resource_count(dev, SYS_RES_IOPORT, 0) == 4)
fdc->port_off = -2;
fdc->flags |= FDC_ISPNP;