performance hit, and on an 80386 processor, it most certainly is. Pull it
from the GENERIC (and DISKLESS "generic") kernels--configure it in
yourself if you actually need it.
16M, thus eating up all candidate bounce pages, which are not allocated
until autoconfiguration runs, by deferring allocation of the buffer
cache pages until after autoconfiguration has run.
XXX This is safe because the i386 port doesn't read any disklabels
XXX during autoconfiguration.
ourselves, but this can still happen on a 386 in copyout & friends.
Fixes pmap_changebit panics seen on 386s. Thanks to Juergen Hannken-Illjes
for providing detailed info on bug occurances.
"struct cpu_cpuid_family", so that we can have a cpu_setup function per
known cpu type. For now use it only for cyrix 6x86 to enable suspend-on-halt
and implement cyrix's workaround to the "coma bug".
memory-mapped device access (manifested by an apparent wedging of the
PCI bus):
- If cpu_class == CPUCLASS_486, disable memory-mapped PCI device access,
leaving only i/o-mapped access enabled.
- Provide a patchable kernel variable (i486_pci_mem_enabled) and a kernel
option (I486_PCI_MEM_ENABLED) to re-enable it.
* Call gdt_init() earlier on.
* Don't bother to check for TSS and LDT descriptors in verr_gdt().
* Nuke the sti special case for trace and breakpoint traps.
* Move some variable declarations into more appropriate locations.
- fix _C_LABEL so that it actually works.
- make __RENAME use _C_LABEL.
- fix __RENAME so that it expects an unquoted argument.
- fix __indr_reference and __warn_references so that they
supply their own final semicolon.
- define __warn_references to nothing if not GNU C (required
by the way it's used).
The __warn_references semicolon change has to be made
so that __warn_references can be defined into nothing.
(A ; all by itself isn't a great idea.) The __indr_reference
change was made for consistency.
Instead, print a diagnostic and return. (Some drivers do this already.)
Also, normalize the diagnostic message, and fix some places where the
printfs were getting ugly.
further. Devices are grouped by types and within the types segregated
by bus. Lots of comments fixed up, lots of nits cleaned. Extensive
comments from Jason and Charles incorporated.
We also really need an automatic way to generate INSTALL from GENERIC,
but that's a project for another day...
finish up incomplete job of moving structure used by bioscall() to
<machine/bioscall.h>, with associated changes in include file strategy,
genassym stuff, etc.
Implement new ISA interface functions:
* isa_intr_alloc() - allocate an ISA IRQ
* isa_mem_alloc() - allocate ISA memory space
* isa_mem_free() - free ISA memory space
pseudo-device rnd # /dev/random and in-kernel generator
in config files.
o Add declaration to all architectures.
o Clean up copyright message in rnd.c, rnd.h, and rndpool.c to include
that this code is derived in part from Ted Tyso's linux code.
Only assembly version for i386 bswap16 and bswap32 for now (bswap64 uses
bswap32). Contribution of assembly versions of these are welcome.
Add byte-swapping of ext2fs metadata for big-endian systems.
Tested on i386 and sparc.
while "findroot()". This happens if a disk without BSD partition is
present, or if the partition does not exist in the disklabel - normal
operation, not worth reporting. (Split out a part to keep indentation
level limited.)
Allow to set "nfs_boot_rfc951" by a conditional to make bootp easier
to use.
tsleep() to be called from an interrupt handler.
The semantics of the scroll-lock key is changed now: it issues a ^S or
^Q, depending on the current state. (It should probably issue
tp->t_cc[VSTOP] or tp->t_cc[VSTART] instead, but this would require more
serious structural changes because there is not always a tty context
present.)
The "scroll lock" LED is now controlled by pcstart()/pcstop(), so it
will show the real state even if the start/stop characters are remapped
or the normal ^S/^Q are used.
by bus type / address (for diskless boot)
-add some (experimental, commented out) code which could be used for
identification of boot disks - should help for BIOS remappings
The former "cyloffset" argument is used as an (indirect) pointer to the
information structures. "cyloffset" has always been set to zero in the
past, so this should be save to do.
-add a function ("lookup_bootinfo") to extract data
-use passed console information, if present, for console attachment
-arrange definitions for console and KGDB device a bit nicer
msgbuf. Note that old 'dmesg' and 'syslogd' binaries will continue running,
though old 'dmesg' binaries will output a few bytes of junk at the start of
the buffer, and will miss a few bytes at the end of the buffer.
-optionally, use Martin's direct hardware access to serial lines
-optionally, new console autoselection method: select serial console
if a character is seen within 1 second
-move awaitkey() here, it can be shared by different bootloaders and
it is used by console autoselection
-collect information about console selection in a "bootinfo" defined
structure, for passing to the kernel
separated "nfswrapper.c"/"nfs.c".
This nfs.c differs from the stand/lib/libsa one in that it does not
have a separate mount() function - the mount is done in open().
* support chip clocks != COM_FREQ, by introducing sc_frequency (for the
mainline code) and adding a frequency parameter right after the rate
parameter to comcnattach() and com_kgdb_attach().
- Make com_isa and com_multi initialize sc_frequency to COM_FREQ.
- Make i386/machdep.c and alpha/dec_xxx.c call com*attach() with the freq.
parameter.
* supio_attach_args get two more fields: a sc_ipl and a sc_arg, both ints.
- com_supio uses the first for interupt establishment (all childs will, as
soon as they exist) and the 2nd for sc_frequency.
- drsupio passes sc_ipl alway as 5, and for the "com"s, sc_arg as 16*115200
- hyper will pass sc_ipl as 6, and sc_arg as 16 * 460800
(currently only CD-ROM drives on i386). The sys/dev/scsipi system provides 2
busses to which devices can attach (scsibus and atapibus). This needed to
change some include files and structure names in the low level scsi drivers.
two DMA channels to do e.g. full-duplex. This allows
a way of specifying the second channel in a sane way.
THIS IS TEMPORARY. The drq2 locator will go away when
the locator system has been changed to allow multiple
values per locator.
device memory mapping cacheable/non-cacheable:
- PG_N doesn't exist on the 386.
- pmap_changebit() only deals with managed pages.
Basically, calling it is unnecessary, and never did anything. Pointed
out by Chuck Cranor, and further discussed with Frank van der Linden.
Also, add a comment about why we don't mark pages non-cacheable in
_bus_dmamem_map().
Delay setting of cn_tab->cn_dev until autoconfiguration attach
to get the minor number right. (not really needed, but for
consistency with pccons and serial consoles)
Comment out unused pccnprobe() and pccninit().
XXX I didn't use "#if PCVT_NETBSD > ???" anymore, nor separated from
FreeBSD. The driver is not centrally maintained anyway.
* make map and alloc take 'flags' rather than 'cacheable,' for
more flexibility.
* rename BUS_BARRIER_* to BUS_SPACE_BARRIER_*, for consistency.
* rename bus_space_copy_* to bus_space_copy_region_* and make their
defns match the updated spec.
Backward compatibility is provided by defining __BUS_SPACE_COMPAT_OLDDEFS,
which is currently defined by default.
anymore). This makes the code cleaner (no more bus tag guessing in the
serial console driver) and it allows to share code with KGDB port
initialization.
The console device is now given as string ("pc", "com0".."com3").
Modelled after alpha port.
keys on a pccons console keyboard.
submitted in PR 899 by Alistair G. Crooks
Note that I only did this for the US type keyboard maps.
This and all other such options should be documented, and perhaps
rennamed with consistant PCCONS_ prefixes.
o option DISPLAY_ISO8859 enables the display of iso-latin1
character set (instead of the IBM page code 437)
o option FRENCH_KBD, GERMAN_KBD or NORVEGIAN_KBD implement
support for national keyboards (implies DISPLAY_ISO8859).
Originally supplied in PR #1529
2) Add option PCCONS_REAL_BS which (for US keyboards only) forces
backspace to really be backspace and not delete. Intended to close
PR #2264 submitted by Greg Woods. He wanted it changed for everyone
-- I thought adding a kernel compile option was friendlier.
Note: Both of these sets of options really should be documented in an
i386 specific version of options(4).
get APM debugging code compiled in. also, redo the way APM debugging
printfs are invoked, making sure they're always compiled in if debugging
is enabled, but printing them only if the appropriate bits in the
'apmdebug' global variable are set.
* clean up function prototypes (alphabetize in each category, get rid
of 'STATIC' and make them always static since kernels shouldn't have
static symbols stripped anyway, and debugging kernels definitely don't),
group global variables together and note that they're used as globals).
* Export the following as option-settable and patchable global variables:
apmdebug flags word that controls which APM debugging
printfs are printed, set with APMDEBUG_VALUE.
apm_enabled boolean that indicates whether or not APM should
ever be configured on the machine. true by
default, false if APM_DISABLE defined when
compiling. this option exists so that users
whose systems crash when APM is enabled can
disable it with a patchable kernel variable, and
therefore don't necessarily have to compile
a custom kernel as often. With this option set,
a check is still done to see if the APM BIOS
exists; this variable should be used to prevent
the kernel from ever attaching to the APM BIOS.
apm_force_64k_segments
boolean that tells segment setup code to always
use segments with 64k length. false by default,
true if APM_FORCE_64K_SEGMENTS defined when
compiling. used to just be a compile-time option.
apm_do_idle boolean that controls whether or not the APM code
will try to signal the BIOS that the system is
idle when in the idle loop. true by default,
false if APM_NO_IDLE defined when compiling.
apm_do_standby boolean that controls whether or not standby
requests are ever attempted. true by default,
false if APM_NO_STANDBY defined when compiling.
used to just be a compile-time option.
apm_v11_enabled boolean that controls whether or not the driver
attempts to use APM spec v1.1 features. If true,
the driver does (it implements v1.1 of the spec),
and if false the driver only uses v1.0 features.
true by default, false if APM_V10_ONLY is defined
when compiling. Really, the driver should be
updated to do APM v1.2 and this option should
either be updated or go the way of the dodo.
used to just be a compile-time option.
apm_bogus_bios (patchable only) controls a few of the APM segment
setup features. (probably should just be a global
and not patchable.)
* make apm_perror print out the full contents of %ax rather than
just the error code, in addition to printing out the string corresponding
to the error code.
* print estimated remaining battery life time as "[<hours>h ]<minutes>m",
which is actually easily understood, rather than as "<hours>:<minutes>
minutes", which is not.
* restructure match and attach so that they work as they should, remove
redundancy, support attaching the APM device but not actually enabling it
so that APM is attached on systems which have it but on which it's broken
in some fatal way.
* add more debugging printfs in the apm control device entry points.
* return EOPNOTSUPP when trying to do a standby ioctl if standbys have
been disabled, rather than returning the less-intuitive ENOTTY.
* rename a few functions for consistency: apmprobe -> apmmatch,
apm_err_translate -> apm_strerror.
* mark a few serious or annoying bogons XXX so people might look at
them later.
* KNF (lots of changes!)
Despite this substantial amount of work, this driver still needs a _lot_ of
work. In particular, the user-land device interface needs to be gutted, and
better support for APM-aware applications and device drivers (i.e. call-backs)
needs to be added (e.g. callback interfaces somewhat like those suggested
by the APM spec).
There are still outstanding known problems with some machines. The true
source of those problems is not currently known, and could be either the APM
BIOSes of the machines in question or this APM driver. However, even if the
problems are caused by APM BIOS bugs, they could be more gracefully worked
around in this driver.
failure. The latter usually would end up being the (nonzero) opcode
placed in %ax before the call, but could conceivably be modified by the
call to be zero. In any case, 0/1 is much nicer than 0/random-nonzero,
as well as being safer.
a global variable (yech, but i'm not out to save the world) which had
the same meaning and ended up being stealthily set in pretty much
the same place.
doing 'if (tag == ...) else if (tag == ...) else panic' rather than
by doing a switch. This makes life easier for people who need
the i386 bus_space_* functions to access spaces other than the normal
memory and I/O spaces, because it allows them to define an i386
bus_space_tag_t as a pointer to a function table, rather than just
an int. (They could define it as an int and cast it to a funtion
table pointer, but I think that's bad karma, and this change doesn't
hurt.)
mark all of 0 -> IOM_BEGIN as used, even though there are some regions
which are left for use by the BIOS (namely, the first 4k page and the
area after biosbasemem*1024 but before IOM_BEGIN). Drivers wishing to
manipulate these areas should map them specially, with _i386_memio_map()
rather than bus_space_map() or i386_memio_map().
disabled even if it's attached. If disabled, ENXIO on open.
(2) in the case where the code32 segment len overflows the I/O hole,
instead of giving up truncate it. (In other words, revert
change (2) in rev 1.21 which i suggested mistakenly.)
(3) map bios data space in the 0->640k range with _i386_memio_map()
rather than with bus_space_map(), so that no accounting checks
are done. The checks which are done to make sure that allocation
in this range is safe are sufficient.
(4) check the return value from _i386_memio_map(), and if it indicates
an error disconnect from the APM BIOS, print an error message,
and return without having marked the device 'enabled'.
functions (which only work on memory and i/o space) to i386_memio_*,
and make bus_space_* in bus.h be #defines which invoke them. This makes
life easier for people who need to define the all of the bus_space functions
so that they work on spaces other than memory and I/O space. Also, add
an _i386_memio_map function which is like i386_memio_map but doesn't do
the extent map checking or allocation. _i386_memio_map and i386_memio_*
are for use only by machine-dependent code.
code for this was in my private tree at one time, and I must have
forgotten to g/c the constant from trap.h). Reported by
enami tsugutomo <enami@but-b.or.jp>, PR #3845.
("filename" must be processed by parsebootfile() first).
Being here:
- remove the special "#ifdef MATTHIAS" mainloop
- make some "booting..." output match reality
- print '\n' and '\b' more controlled
the 'a' partition. Sanity checked by thorpej.
This, incidently, may have been causing user errors in initializing
new disks, as described in (now closed) pr-2729 from Scott Reynolds,
because users typically don't try to edit their c partitions to be
"correct".
drochner