CVS log for src/sys/arch/mac68k/dev/sbc.c

Up to [cvs.NetBSD.org] / src / sys / arch / mac68k / dev

Request diff between arbitrary revisions

Keyword substitution: kv
Default branch: MAIN

Remove CHECKMORE.

This was a performance penalty and was found to be unnecessary as the root
cause of errors doing pdma interrupt xfers was addressed in an earlier commit.

Serialize pdma xfers.

The same is done for pdma interrupt xfers.

Make writing of dummy byte conditional.

This is insurance against devices that do not leave ACK hanging at the end of
each transfer.

Final part of PR kern/58452.

This delay is no longer necessary.

The delays throught the code were there to ensure error free transfers.  This
has been fixed in an eariler commit today by serializing the drq interrupt.

Part of PR kern/58452.

No need any additional delay waiting for DREQ.

This results in utilites like time(1) reporting results closer to reality
when timing the transfer of data.  With the added benefit that transfers are
slightly faster.

Part of PR kern/58452.

Serialize drq interrupt.

This results in error free transfers between disks.

Part of PR kern/58452.

Spinkle a little delay.

This was needed to ensure a reliable transfer of data.

Write an extra byte to handle the last ACK from 5380.

This was found from the example code in the NCR5380 Interface Manual.

This is not needed (but does not hurt) on the 53C80 and later.

This fixed many problems from stuck busy to phase mismatches when aborting
transfers and avoiding hard lockups of the scsi subsystem originating from
what I first thought were errors from the target device.

XXX pullup-9, pullup 10.

I will look next at the pdma_out function to do the same there.

Rework of previous.

While here set the priority in sync witho pdma_in/out.
Use sbc_ready same as pdma_in/out.

Use smaller size when sending PDMA data, and check that the next block can
be sent before sending it.

This fixes PDMA support (which previously has been flakey at best).

These changes provide maximum performance and importantly reliablity.

The lack of support for the PDMA intr should now be revisited on the PowerBook
5xx series (Unfortunately I don't have the hw to test with).

These changes have been tested on emulated (PiSCSI/BlueSCSI) and real hw.

Introduce a flag to disable PDMA when writing.

The use of pdma when writing would always cause a panic on my Powerbook 160,
possibly others as well.

As posted to tech-kern.

Honor the XS_CTL_POLL flag and fallback to polled io instead of PDMA.

As posted to tech-kern.

Sync with HEAD

Sync with HEAD

Resolve a couple of conflicts (result of the uimin/uimax changes)

Rename min/max -> uimin/uimax for better honesty.

These functions are defined on unsigned int.  The generic name
min/max should not silently truncate to 32 bits on 64-bit systems.
This is purely a name change -- no functional change intended.

HOWEVER!  Some subsystems have

	#define min(a, b)	((a) < (b) ? (a) : (b))
	#define max(a, b)	((a) > (b) ? (a) : (b))

even though our standard name for that is MIN/MAX.  Although these
may invite multiple evaluation bugs, these do _not_ cause integer
truncation.

To avoid `fixing' these cases, I first changed the name in libkern,
and then compile-tested every file where min/max occurred in order to
confirm that it failed -- and thus confirm that nothing shadowed
min/max -- before changing it.

I have left a handful of bootloaders that are too annoying to
compile-test, and some dead code:

cobalt ews4800mips hp300 hppa ia64 luna68k vax
acorn32/if_ie.c (not included in any kernels)
macppc/if_gm.c (superseded by gem(4))

It should be easy to fix the fallout once identified -- this way of
doing things fails safe, and the goal here, after all, is to _avoid_
silent integer truncations, not introduce them.

Maybe one day we can reintroduce min/max as type-generic things that
never silently truncate.  But we should avoid doing that for a while,
so that existing code has a chance to be detected by the compiler for
conversion to uimin/uimax without changing the semantics until we can
properly audit it all.  (Who knows, maybe in some cases integer
truncation is actually intended!)

Rebase to HEAD as of a few days ago.

Rebase.

Remove an unused variable

resync with head

sync with (a bit old) head

Remove the R1 syntactic sugar, since it collides with a #define in
<m68k/regs.h>, breaking the build.

The R1s serve to bring the buffer pointer to a 4 byte boundary, but
that should be clear from the context.

sync with head

Use lwp_getpcb() on m68k ports, clean from struct user usage.

Sync with HEAD.

Split devict_t/softc for ncr5380sbc SCSI, and misc cosmetic changes.

sync with head.

Sync with HEAD.

Kill caddr_t; there will be some MI fallout, but it will be fixed shortly.

sync with head.

sync with head

sync with head

gcc4 "uninitialized" variable;  previous workaround doesn't satisfy gcc4 -
explicitly initialize drq.

sync with head.

Sync with head.

Merge 2006-05-24 NetBSD-current into the "peter-altq" branch.

sync with head.

sync with head

Fix a bunch of cast lvalues.

sync with head.

Pull up following revision(s) (requested by chs in ticket #1146):
	sys/arch/mac68k/dev/sbc.c: revision 1.49
add fault-protection in sbc_pdma_in() like in all the other PDMA functions.
fixes PR 9679.

add fault-protection in sbc_pdma_in() like in all the other PDMA functions.
fixes PR 9679.

__asm__ -> __asm
__const__ -> const
__inline__ -> inline
__volatile__ -> volatile

merge ktrace-lwp.

Sync with HEAD. Here we go again...

Fixes for volatile problems

Sync with HEAD.

de-__P, remove register, ansify, b* -> mem*.

Fix the sync with head I botched.

Sync with HEAD.

Sync with HEAD

__KERNEL_RCSID()

DMA, not dma nor Dma.

Catch up to -current.

sync kqueue with -current; this includes merge of gehenna-devsw branch,
merge of i386 MP branch, and part of autoconf rototil work

remove trailing \n in panic().  approved perry.

Update thorpej_scsipi to -current as of a month ago
A i386 GENERIC kernel compiles without the siop, ahc and bha drivers
(will be updated later). i386 IDE/ATAPI and ncr work, as well as
sparc/esp_sbus. alpha should work as well (untested yet).
siop, ahc and bha will be updated once I've updated the branch to current
-current, as well as machine-dependant code.

Add a ncr5380_attach() routine which does part of the initialization, attaches
the scsibus, and does the addref/delref dance.

Pull up to last week's -current.

Pull up the version from the trunk; some changes crept in that ought not
to have been committed.

Sync with -current

[Redo] Sync with main branch.

[Redo] Catch up with recent changes to scsipi flags.  Closes PR 8675.

Centralize the declaration and clearing of `cold'.

Sync with main branch.

Adapt to the new scsipi_adapter interface.

Garbage-collect the open_target_lu and close_target_lu entry points from
struct scsipi_adapter; they were not used.

Add a scsipi_ioctl entry point to struct scsipi_adapter.  This will be
used to issue ioctl commands to the host adapters.

Inspired by PR #6090, from Matt Jacob.

defopt DDB.

Pull up 1.34 (scottr)

We haven't needed a driver-specific minphys() for quite some time now.

Update for changes to config.

Sync w/ trunk: fix potential hang during autoconfig.  (scottr)

PR 4078:  Enabling the SCSI IRQ interrupt to allow reselects to work
causes the MI interrupt handler to barf when we get a 5380 RST interrupt
while probing.  Worse, the VIA latches the interrupt, so simply having
all interrupts disabled during autoconfig doesn't resolve the problem.
[I demonstrated the latter on a IIci, which erroneously reports a
reselection attempt(!) after autoconfig is complete.  The latched
interrupt results from the SCSI bus reset we do when initializing the
bus.]

Since interrupts must be enabled during autoconfig anyway (sigh), test
to see if autoconfig has completed in sbc_irq_intr().  If not, we don't
pass the interrupt up to the MI interrupt handler.  Also, make sure to
clear the VIA interrupt if we're servicing an unclaimed 5380 RST
interrupt.

Thanks to Bill Studenmund for providing the key insight needed to unlock
this problem.

DEBUG -> SBC_DEBUG (scottr)

Undo parts of 1.23 to fix autoconfig hangs on some Macs.   Not on trunk - trunk version is more correct.   (scottr)

Update marc-pcmcia branch from trunk.

Remove advertising clause.

Update marc-pcmcia branch from trunk.

Catch bus error in sbc_pdma_out() and panic with a reasonable
error message.

Update marc-pcmcia branch from trunk.

Merge scsipi branch in the mainline. This add support for ATAPI devices
(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.

Update marc-pcmcia branch from trunk.

Sync with trunk.

Make #include directives consistent.  All MD headers are of the form:

	#include <mac68k/{dev,mac68k}/foo.h>

Updates for new scsipi subsystem. Actally known to work on i386 and sparc.

Due to reliability problems on some models, back out the more radical of
the pc532/dev/ncr.c-inspired changes; in particular, wait for previous
transfer to complete before starting another, as we used to do.  Retain
splbio() protection and simplified sbc_pdma_in().

Update for mac68k_buserr_addr -> m68k_fault_addr change.

Tweak sbc_drq_intr() (interrupt-driven PDMA interrupt handler); it
now works properly on at least some disks.  Further testing is
necessary.

Undo a hack added in the last overhaul (which caused an extraneous byte
to be written to the target).  Also, use the sc_min_dma_len field
to determine whether to use PDMA or PIO.

For small writes, use PIO rather than PDMA; this increases the reliability
of the driver (considerably, on some systems/targets).

Rework PDMA read/write functions based on Matthias Pfaller's pc532 ncr
driver.  Major highlights:

 - Protect with splbio(), now that interrupts from the chip might be
   enabled by default;
 - Wait for current transfer to complete before returning, rather than
   wait for previous transfer to complete before starting another;
 - Instead of logging a "timeout" message when a target disconnects,
   just return, letting the sbc_ready() and sbc_wait_not_req() print
   timeout messages if such really did occur.

With interrupt changes to sbc_obio.c, disconnect/reselect now works if
configured into the kernel.

Merge in changes from The Trunk, partially just reimplementing newarp.

Split bus-dependent autoconfig code from the driver.  This allows for
the future addition of a Duo Dock front end.

Convert to generalized VIA interrupt registration

Minimal changes to adapt to removal of NCR5380_PERMIT_RESELECT.
This driver should be updated to allow per-target control over
disconnect/reselect, but I will leave that to the port masters.

Update copyright and license.

#ifdef out write hack, and fix possible uninitialized variable.

Fix typo in last:  cf->cf_unit -> sc->dv.dv_unit

Add basic support for a Duo dock SCSI controller, from Daishi Kato.

XXX - This could probably be improved if all docking hardware has a
NuBus declaration ROM that we can get to.  This should be investigated
further.

Convert all foo_match() functions to use a `struct cfdata *' for their
second argument.  The NuBus autoconfig code had to be reorganized as a
result of this, and looks much more like a directly-attached bus now.
These changes eliminate __BROKEN_INDIRECT_CONFIG.

Use bitmask_snprintf().

backout previous kprintf change

printf -> kprintf, sprintf -> ksprintf

(1) set scsi_link channel to either the appropriate channel (if a
    multi-channel driver), or to SCSI_CHANNEL_ONLY_ONE if a
    single-channel driver.
(2) use scsiprint() rather than a locally-defined autoconfig print
    function, and kill any locally-defined print function.

change cfprint_t type definition to take a const char *, rather than
a char *, because that's what was really intended, and because
if the print function modifies the string, various things could become
unhappy (so the string should _not_ be modified).

Pulled up from revision 1.9 by request of Scott Reynolds.
Updates copyright/licence and comments.  No code changes.

Update some comments, and the copyright

Pull up from trunk:
>Deal with interrupt flags more carefully, and use the correct offset
>for PB500-series SCSI I/O.  While I'm here, update the copyright.

Deal with interrupt flags more carefully, and use the correct offset
for PB500-series SCSI I/O.  While I'm here, update the copyright.

- Simplify writes in sbc_drq_intr().
- When finished writing, if the SCSI bus has BSY asserted, write another
  byte to the SBC to ensure we get an interrupt.
- Unflag SCSI interrupts on the VIA whenever we clear the interrupts
  on the SBC itself.

Remove some gross hacks that were added due to interrupt grossness that
has been partially fixed.  Also, add a new flag (which rearranges them
a bit, unfortunately) to use PDMA for polled transfers.

Prototype for -Wall -Wstrict-prototypes -Wmissing-prototypes -Wno-uninitialized
Also change the device probing scheme to use something a bit more rational.
A current side-effect is that nubus cards are double-mapped.  I expect
to fix that shortly.
Also change splclock() to block everything but serial hardware interrupts.

Kill unused variables and fix prototypes.

- Drop SCSI IRQ interrupts if we are already handling one.  This works around
  a condition that occurs with some slow SCSI devices when they disconnect
  (e.g. the AppleCD 600), generating spurious selection interrupts.
- Reorganize the debugging code slightly.

Ooops.  ncr_sbcreg.h -> sbcreg.h

Renamed these files.