in many device attach paths, allocate memory with M_WAITOK instead of M_NOWAIT
and remove code to handle failures that can no longer happen.

- add or adjust /* FALLTHROUGH */ where appropriate
- add __unreachable() after functions that can return but won't in
this case, and thus can't be marked __dead easily

branches: 1.55.64;
struct device * -> device_t, no functional changes intended.

ansify function definitions

Change about 4500 of the K&R function definitions to ANSI ones.
There are still about 1600 left, but they have ',' or /* ... */
in the actual variable definitions - which my awk script doesn't handle.
There are also many that need () -> (void).
(The script does handle misordered arguments.)

branches: 1.52.4; 1.52.10;
split device/softc
Now I'm through - all my devices are split.

branches: 1.51.4; 1.51.6; 1.51.8;
use aprint_*_dev and device_xname

branches: 1.50.48;
__unused removal on arguments; approved by core.

- sprinkle __unused on function decls.
- fix a couple of unused bugs
- no more -Wno-unused for i386

branches: 1.48.2; 1.48.4;
redo with c99 initializers

- comment out impossible comparison
- add missing initializer

- wrap memspace {}, pointed out by rui.
- comment out the memspace none case, pointed out by rui.
- move declarations to the top to avoid brace mess.

Split the tuple parsing code in its own function for each kind, instead of
having a single parsing function that has more than 1000 lines of code.

Back out rev 1.42 and 1.43. It's been causing much more hassle than it should
and I don't feel like continue arguing about it.

Add misssing }. Sorry for breaking the build.

In pcmcia_parse_cis_tuple(), refactor the memspace section introducing
a switch statement and fixing CID 566.

branches: 1.41.2; 1.41.4; 1.41.6;
When unmapping and remapping, the mem handle and mem tags can change. Do
not assume that they will remain the same.

branches: 1.40.2; 1.40.4; 1.40.6;
merge ktrace-lwp.

branches: 1.39.4;
nuke trailing whitespace

de-__P

branches: 1.37.4; 1.37.6;
Rewrite the loop without using useless SIMPLEQ_FOREACH macro.

deal with PCMCIA_TPCE_IO_RANGE_ADDRSIZE_NONE and
PCMCIA_TPCE_IO_RANGE_LENGTHSIZE_NONE properly. enami
solves kern/26726.

Constify.

Swap around the output in some other cases so it's consistent.
Remove some more DELAY()s.

Only DELAY() once per CIS region, not once per tuple.
For PCMCIACISDEBUG, print out the tuple *before* parsing it or calling the
callback, so that we can see the tuple that the callback claimed.

Typo in comment. From OpenBSD.

branches: 1.31.6;
Fix multi-function card memory problems:
- centralize pcmcia function allocation and free'ing.
- free the cfe too, not just the pf in the multifunction card case.
- don't free pointers while walking the list, because free() will
fill the memory with deadbeef, thus killing list walking.

SIMPLEQ rototill:
- implement SIMPLEQ_REMOVE(head, elm, type, field). whilst it's O(n),
this mirrors the functionality of SLIST_REMOVE() (the other
singly-linked list type) and FreeBSD's STAILQ_REMOVE()
- remove the unnecessary elm arg from SIMPLEQ_REMOVE_HEAD().
this mirrors the functionality of SLIST_REMOVE_HEAD() (the other
singly-linked list type) and FreeBSD's STAILQ_REMOVE_HEAD()
- remove notes about SIMPLEQ not supporting arbitrary element removal
- use SIMPLEQ_FOREACH() instead of home-grown for loops
- use SIMPLEQ_EMPTY() appropriately
- use SIMPLEQ_*() instead of accessing sqh_first,sqh_last,sqe_next directly
- reorder manual page; be consistent about how the types are listed
- other minor cleanups

branches: 1.29.8; 1.29.10;
Call malloc(9) with M_ZERO flag instead of memset() after malloc().

don't need <sys/types.h> when including <sys/param.h>

add RCSIDs

Commit changes I posted to tech-kern on Oct 14, which makes the ethernet part
of my Xircom RealPort Ethernet 10/100 + Modem (REM56G) work.
The modem part is still not usable (this would need some PCMCIA magic that
I don't know how to do; see my message to tech-kern on Oct 07).
Change to pcmcia_cis.c from OpenBSD.
I checked that this doesn't break a 3com 3C562D (ethernet+modem) which I
also have.

branches: 1.25.2;
Add simple boundary check on CIS parsing.
I have a weird PC-card-style appliance (I'm not sure I may call it a PC card)
whose ``CIS'' reads zeros forever, which caused kernel panic.

For your interest, it is a cooling fan to be inserted to a PC card slot.

branches: 1.24.2; 1.24.4;
bzero -> memset

Turn off PCMCIACISDEBUG by default. This should get set with options in a config file

Change place of tuple.{memt,memh} initialization.

branches: 1.21.2;
Stopgap fix for PR 9984: sprinkle some (small) delays at appropriate
places into the CIS reading code.

The card in question has IO8 only enabled in its CIS info and is apparently
not able to keep up with quick reads. It words fine in a pcmcia slot but
panics(!) the kernel in a TI 1250 cardbus slot. This may be a failure of
the pci cardbus code when initializing this bridge. When finding (and
fixing) that, we should back this change out.

The card I am testing with is not broken, I have multiple versions of it
(AVM Fritz! pcmcia ISDN card), all work fine on windows and all cause
us to panic because of bogus CIS info read.

XXX - panicing because of bogus CIS data is probably another error.

Add a cis decoding function which decodes some CISTPL_FUNCE tuples for
disk and network functions.

apply Kenji Aoyama <aoyama@nk.mcu.or.jp>'s patch
on port-hpcmips-ja@jp.netbsd.org Mon, 10 Jul 2000 21:21:47 +0900

FIX:
Can't read CIS tuple chain:some TDK CF Flash memory card.

branches: 1.18.4;
Cosmetic changes.

map common memory 8 bit for cis scan when following longlinks.

The cfe is initialized to the default entry, when a table entry value
is present don't OR in the flags with the defaults, replace them.

Don't overwrite cfe->flags.

Some additional sanity checks to keep this code from bursting into
flame if presented with bad CIS data (e.g. because of a memory space
conflict):
* more sanity checking on LONGLINK_MFC CIS tuples (exact length check, sanity
check on size vs. size of array we allocated for them), to avoid various
missteps which could cause this code to complete trash the kernel stack.
* clear the entire contents of the state structure before processing, so
things like uninitted pointers will actually have a known value!
* be more careful with CISTPL_CFTABLE_ENTRY: check to see that the current
state's default_entry isn't NULL before dereferencing it.

Print range of all iospaces or all memspaces if their length aren't zero.

branches: 1.12.2; 1.12.4; 1.12.8;
In pcmcia_pars_cis_tuple(): in the PCMCIA_TPCE_FS_MEMSPACE_TABLE case
there is one more memspace in the table then the PCMCIA_TPCE_MS_COUNT
mask produces.

In pcmcia_print_cis(): When printing io and mem spaces don't append trailing
0s if length or hostaddr are zero, instead DTRT.

branches: 1.10.2; 1.10.4;
check for end of tuple while reading cfe data

fix version string (from PC Card Standard, April 1998 (Release 6.1))

Use PCMCIA_FUNCTION definitions.
Initialize function ID in new pcmcia_function structures to
PCMCIA_FUNCTION_UNSPEC, not PCMCIA_FUNCTION_MULTIFUNCTION (0).

Add a devlist2h.awk so that we don't sprinkle the same constants over
each device driver file and use it.

better MFC parsing (for 3c56[23]D), and make sure the pf is zeroed out before using it (for adaptec 1460b)

add some information from "PC Card standard March 1997"

- fix pcmcia_scan_cis to return:
-1 error
0 desired tuple not found
1 desired tuple found

Make `manufacturer' and `product' 32bit wide so that they can contain
both 16bit id values found in card CIS and special value to indicate
that no value found in CIS. Use that special value for the card that
doesn't have id values. Test that value in ne2000_match().

Pull marc-pcmcia branch down from trunk.

branches: 1.1.2;
file pcmcia_cis.c was initially added on branch marc-pcmcia.

- add or adjust /* FALLTHROUGH */ where appropriate
- add __unreachable() after functions that can return but won't in
this case, and thus can't be marked __dead easily

struct device * -> device_t, no functional changes intended.

ansify function definitions

Change about 4500 of the K&R function definitions to ANSI ones.
There are still about 1600 left, but they have ',' or /* ... */
in the actual variable definitions - which my awk script doesn't handle.
There are also many that need () -> (void).
(The script does handle misordered arguments.)

branches: 1.52.4; 1.52.10;
split device/softc
Now I'm through - all my devices are split.

branches: 1.51.4; 1.51.6; 1.51.8;
use aprint_*_dev and device_xname

branches: 1.50.48;
__unused removal on arguments; approved by core.

- sprinkle __unused on function decls.
- fix a couple of unused bugs
- no more -Wno-unused for i386

branches: 1.48.2; 1.48.4;
redo with c99 initializers

- comment out impossible comparison
- add missing initializer

- wrap memspace {}, pointed out by rui.
- comment out the memspace none case, pointed out by rui.
- move declarations to the top to avoid brace mess.

Split the tuple parsing code in its own function for each kind, instead of
having a single parsing function that has more than 1000 lines of code.

Back out rev 1.42 and 1.43. It's been causing much more hassle than it should
and I don't feel like continue arguing about it.

Add misssing }. Sorry for breaking the build.

In pcmcia_parse_cis_tuple(), refactor the memspace section introducing
a switch statement and fixing CID 566.

branches: 1.41.2; 1.41.4; 1.41.6;
When unmapping and remapping, the mem handle and mem tags can change. Do
not assume that they will remain the same.

branches: 1.40.2; 1.40.4; 1.40.6;
merge ktrace-lwp.

branches: 1.39.4;
nuke trailing whitespace

de-__P

branches: 1.37.4; 1.37.6;
Rewrite the loop without using useless SIMPLEQ_FOREACH macro.

deal with PCMCIA_TPCE_IO_RANGE_ADDRSIZE_NONE and
PCMCIA_TPCE_IO_RANGE_LENGTHSIZE_NONE properly. enami
solves kern/26726.

Constify.

Swap around the output in some other cases so it's consistent.
Remove some more DELAY()s.

Only DELAY() once per CIS region, not once per tuple.
For PCMCIACISDEBUG, print out the tuple *before* parsing it or calling the
callback, so that we can see the tuple that the callback claimed.

Typo in comment. From OpenBSD.

branches: 1.31.6;
Fix multi-function card memory problems:
- centralize pcmcia function allocation and free'ing.
- free the cfe too, not just the pf in the multifunction card case.
- don't free pointers while walking the list, because free() will
fill the memory with deadbeef, thus killing list walking.

SIMPLEQ rototill:
- implement SIMPLEQ_REMOVE(head, elm, type, field). whilst it's O(n),
this mirrors the functionality of SLIST_REMOVE() (the other
singly-linked list type) and FreeBSD's STAILQ_REMOVE()
- remove the unnecessary elm arg from SIMPLEQ_REMOVE_HEAD().
this mirrors the functionality of SLIST_REMOVE_HEAD() (the other
singly-linked list type) and FreeBSD's STAILQ_REMOVE_HEAD()
- remove notes about SIMPLEQ not supporting arbitrary element removal
- use SIMPLEQ_FOREACH() instead of home-grown for loops
- use SIMPLEQ_EMPTY() appropriately
- use SIMPLEQ_*() instead of accessing sqh_first,sqh_last,sqe_next directly
- reorder manual page; be consistent about how the types are listed
- other minor cleanups

branches: 1.29.8; 1.29.10;
Call malloc(9) with M_ZERO flag instead of memset() after malloc().

don't need <sys/types.h> when including <sys/param.h>

add RCSIDs

Commit changes I posted to tech-kern on Oct 14, which makes the ethernet part
of my Xircom RealPort Ethernet 10/100 + Modem (REM56G) work.
The modem part is still not usable (this would need some PCMCIA magic that
I don't know how to do; see my message to tech-kern on Oct 07).
Change to pcmcia_cis.c from OpenBSD.
I checked that this doesn't break a 3com 3C562D (ethernet+modem) which I
also have.

branches: 1.25.2;
Add simple boundary check on CIS parsing.
I have a weird PC-card-style appliance (I'm not sure I may call it a PC card)
whose ``CIS'' reads zeros forever, which caused kernel panic.

For your interest, it is a cooling fan to be inserted to a PC card slot.

branches: 1.24.2; 1.24.4;
bzero -> memset

Turn off PCMCIACISDEBUG by default. This should get set with options in a config file

Change place of tuple.{memt,memh} initialization.

branches: 1.21.2;
Stopgap fix for PR 9984: sprinkle some (small) delays at appropriate
places into the CIS reading code.

The card in question has IO8 only enabled in its CIS info and is apparently
not able to keep up with quick reads. It words fine in a pcmcia slot but
panics(!) the kernel in a TI 1250 cardbus slot. This may be a failure of
the pci cardbus code when initializing this bridge. When finding (and
fixing) that, we should back this change out.

The card I am testing with is not broken, I have multiple versions of it
(AVM Fritz! pcmcia ISDN card), all work fine on windows and all cause
us to panic because of bogus CIS info read.

XXX - panicing because of bogus CIS data is probably another error.

Add a cis decoding function which decodes some CISTPL_FUNCE tuples for
disk and network functions.

apply Kenji Aoyama <aoyama@nk.mcu.or.jp>'s patch
on port-hpcmips-ja@jp.netbsd.org Mon, 10 Jul 2000 21:21:47 +0900

FIX:
Can't read CIS tuple chain:some TDK CF Flash memory card.

branches: 1.18.4;
Cosmetic changes.

map common memory 8 bit for cis scan when following longlinks.

The cfe is initialized to the default entry, when a table entry value
is present don't OR in the flags with the defaults, replace them.

Don't overwrite cfe->flags.

Some additional sanity checks to keep this code from bursting into
flame if presented with bad CIS data (e.g. because of a memory space
conflict):
* more sanity checking on LONGLINK_MFC CIS tuples (exact length check, sanity
check on size vs. size of array we allocated for them), to avoid various
missteps which could cause this code to complete trash the kernel stack.
* clear the entire contents of the state structure before processing, so
things like uninitted pointers will actually have a known value!
* be more careful with CISTPL_CFTABLE_ENTRY: check to see that the current
state's default_entry isn't NULL before dereferencing it.

Print range of all iospaces or all memspaces if their length aren't zero.

branches: 1.12.2; 1.12.4; 1.12.8;
In pcmcia_pars_cis_tuple(): in the PCMCIA_TPCE_FS_MEMSPACE_TABLE case
there is one more memspace in the table then the PCMCIA_TPCE_MS_COUNT
mask produces.

In pcmcia_print_cis(): When printing io and mem spaces don't append trailing
0s if length or hostaddr are zero, instead DTRT.

branches: 1.10.2; 1.10.4;
check for end of tuple while reading cfe data

fix version string (from PC Card Standard, April 1998 (Release 6.1))

Use PCMCIA_FUNCTION definitions.
Initialize function ID in new pcmcia_function structures to
PCMCIA_FUNCTION_UNSPEC, not PCMCIA_FUNCTION_MULTIFUNCTION (0).

Add a devlist2h.awk so that we don't sprinkle the same constants over
each device driver file and use it.

better MFC parsing (for 3c56[23]D), and make sure the pf is zeroed out before using it (for adaptec 1460b)

add some information from "PC Card standard March 1997"

- fix pcmcia_scan_cis to return:
-1 error
0 desired tuple not found
1 desired tuple found

Make `manufacturer' and `product' 32bit wide so that they can contain
both 16bit id values found in card CIS and special value to indicate
that no value found in CIS. Use that special value for the card that
doesn't have id values. Test that value in ne2000_match().

Pull marc-pcmcia branch down from trunk.

branches: 1.1.2;
file pcmcia_cis.c was initially added on branch marc-pcmcia.