I've fixed the problem that's been keeping me from using anything newer than
1.4.1 now. I tracked down the problem checkin, it's the big reorg of
nextdma.c between 1.19 and 1.20. It didn't introduce a new bug but it
activated a check which wasn't activated before. It's the
is-the-limit-in-the-right-window-check which was ifdeffed before because
some DMA-regs would sometimes have strange values. I think I've fixed the
DMA-reg stuff for now: at the end of nextdma_intr, when the csr is poked
to make DMA do something, I think the check for the ENABLE bit introduces a
race condition. I fixed this by unconditionally setting DMACSR_SETENABLE,
this seems to work and also makes the code more readable. I've also tried
setting DMACSR_SETSUPDATE unconditionally and this also works well, but I
don't know what it implies. Unless you have reasons to not set SUPDATE all
the time, I'd suggest making this change as well, it makes the code cleaner
and faster...
I've tested this patch and it does stop the panics, although I don't think setting
SUPDATE all the tima as he suggests is a good idea. The "SUPDATE" bit implies
a single update (i.e. the end of a dma chain.)
"To fully support self-modifying code in any situation, it is imperative that
a CPUSHA intrcution is executed before the execution of the first self-modified
instruction. The CPUSHA instruction has the effect of ensuring that there is
no stale data iin memory, the pipeline is flushed, and instruction prefetches
are repeated and taken from external memory."
I verified that this is the only way (I can think of) to make the sigtramp
regression test work on 68040. doing cpushl dc; cinvl ic; over the affected
address range, then nop (to synchronize the pipeline) is not enough; apparently
the nop does not FLUSH the pipeline and prefetch...
Note that the 68060 UM has copied the above cited passage, but in fact this is
not true. This might be connected to the fact that the 68060 does ensure
memory access order under most conditions.
remove GENERIC.v6 file (as it is part of GENERIC now).
"faith" interface is commented out by default as it is not really for
general use.
IPsec items are commented out as well, though we can enable "options IPSEC"
without export-related issue ("options IPSEC" will enable authentication
portion only). We may need to think about it again.
if you have problem compiling with INET6 on archs I do not have access to,
please contact me.
XXX what to do with arch/arm32/SHARK{,.v6}?
now uses the DMACSR_READ bit and no longer keeps _nd_dmadir in softc
unified transfer cleanup code, now in routine next_dma_finish_xfer()
fixed bounds checking on registers after transfer.
removed checking for bus errors since the bit is always set on some nexts
(specifically, on mourning, a 25mhz 68040 mono slab)
fixed a couple of dma bugs involving chaining dma buffers.
DMACSR_READ is now a CSR status bit which can be used to know if current transfer is
from cpu to device.
the old DMACSR_READ bit is renamed DMACSR_SETREAD. This is a control bit that tells
the dma transfer to be from cpu to device.
that is priority is rasied. Add a new spllowersoftclock() to provide the
atomic drop-to-softclock semantics that the old splsoftclock() provided,
and update calls accordingly.
This fixes a problem with using the "rnd" pseudo-device from within
interrupt context to extract random data (e.g. from within the softnet
interrupt) where doing so would incorrectly unblock interrupts (causing
all sorts of lossage).
XXX 4 platforms do not have priority-raising capability: newsmips, sparc,
XXX sparc64, and VAX. This platforms still have this bug until their
XXX spl*() functions are fixed.
- fix emitrules() like emitfiles() to deal with the prefix (otherwise it
would attempt to find the file in the normal base for the NORMAL_C rule).
- add emitincludes() which adds include directives for each prefix to the
$INCLUDES variable in the makefile.
- add %INCLUDES to each Makefile.arch to deal with the above.
this makes "prefix" actually work in a usable manner, and now i can move
on to fixing compiler warnings (errors) in the ESP code. :)
changed most of the spl* to use _splraise to match current usage
this fixes an observed bug where zstty_attach was calling splzs causing
interrupts to be turned on prematurely.
- remove "need-flag" for mac68k esp driver, as it is not used in anywhere
and conflicts with IPsec ESP header.
This should be the only MD change in IPv6 support, except kernel config file.
Very sorry if you have any compilation problem with it (I believe it is okay).
If your favorite arch is not included in here, please add a
call to ip6intr() from softintr handle.
return 0 instead of EINVAL. In this case, there will only be one media
type, and the upper-level if_media code will ensure that the user is
setting to that type.
This matches changes Jason made a while back to sys/dev/ic/lance.c
some internal reference counting on PT pages. We still allocate them
with the page fault routine (a wire-fault, now), but no longer free
PT pages from pmap_pageable().
has PAGEABLE and INTRSAFE flags. PAGEABLE now really means "pageable",
not "allocate vm_map_entry's from non-static pool", so update all map
creations to reflect that. INTRSAFE maps are maps that are used in
interrupt context (e.g. kmem_map, mb_map), and thus use the static
map entry pool (XXX as does kernel_map, for now). This will eventually
change now these maps are locked, as well.
managed pages, into KVA space. Since the pages are managed, we should
use pmap_enter(), not pmap_kenter_pa().
Also, when entering the mappings, enter with an access_type of
VM_PROT_READ | VM_PROT_WRITE. We do this for a couple of reasons:
(1) On systems that have H/W mod/ref attributes, the hardware
may not be able to track mod/ref done by a bus master.
(2) On systems that have to do mod/ref emulation, this prevents
a mod/ref page fault from potentially happening while in an
interrupt context, which can be problematic.
This latter change is fairly important if we ever want to be able to
transfer DMA-safe memory pages to anonymous memory objects; we will need
to know that the pages are modified, or else data could be lost!
Note that while the pages are unowned (i.e. "just DMA-safe memory pages"),
they won't consume any swap resources, as the mappings are wired, and
the pages aren't on the active or inactive queues.
directly, call the function pointer (*if_input)(ifp, m). The input routine
expects the packet header to be at the head of the packet, and will adjust
as necessary. Privatize the layer 2 input and output routines, allowing
*_ifattach() to set them up as appropriate.
the child inherits the stack pointer from the parent (traditional
behavior). Like the signal stack, the stack area is secified as
a low address and a size; machine-dependent code accounts for stack
direction.
This is required for clone(2).
define a flag UVM_PGA_USERESERVE to allow non-kernel object
allocations to use pages from the reserve.
use the new flag for allocations in pmap modules.
When we put a page on the collection list, we must subtract NPVPPG from the
total free count: one for each pv_entry that's free in that page, and one for
each free pv_entry in other pages that we're going to eat by moving the ones
in the page being collected.
* Map the message buffer with access_type = VM_PROT_READ|VM_PROT_WRITE `just
because'.
* Map the file system buffers with access_type = VM_PROT_READ|VM_PROT_WRITE to
avoid possible problems with pagemove().
* Do not use VM_PROT_EXEC with either of the above.
* Map pages for /dev/mem with access_type = prot. Also, DO NOT use
pmap_kenter() for this, as we DO NOT want to lose modification information.
* Map pages in dumpsys() with VM_PROT_READ.
* Map pages in m68k mappedcopyin()/mappedcopyout() and writeback() with
access_type = prot.
* For now, bus_dma*(), pmap_map(), vmapbuf(), and similar functions still use
access_type = 0. This should probably be revisited.
These patches include:
Added framebuffer mapping for color framebuffer to support color wscons
which is coming soon.
Renamed wskbdmap_mfii[ch] to wskbdmap_next[ch]
Changed video to be white on black instead of black on white.
Now handles and discards mouse interrupts.
Video and keyboard is now working on mono machines.
write operations appear to lose scsi interrupts and causes timeouts.
changes in this checkin include:
a nextdma bugfix causing a diagnostic check to erroneously trigger
Changed tail strategy to only use tail buffer for the minimal end slop
changed expected write dma overrun to 32 bytes.
turned on nextdma diagnostic check for dma end address since it
no longer gets triggerred by ethernet dma and helps debug scsi dma.
Added esp debugging printout and support.
a filesystem, but is not stable enough yet for general use.
increased priority of ethernet interrupts, mostly useful to aid debugging
of scsi interrupts while using an nfs disk.
added additional debugging output in the next dma driver.
perform extranneous cache flushes/purges before dma reads/ after dma writes
to aid debugging of scsi dma.
Make the initializer for BAUDLO depend on PCLK directly; it was incorrect on
some ports where PCLK is not 4.9152MHz.
XXX Is the default value actually used?
it still probes the disks ok, but is having trouble with 0xdeadbeef
in buffers read from the disk. Suspected problems with DMA alignment
and toggling the secret ninja dma control register
remove check for normal chaining interrupts and add check for combined chaining interrupts
only when used with a tail chain. This allows scsi to work for short unaligned buffers
and large aligned buffers.
This fix removes putting 0xfeedbeef in the unused restart registers.
When that was done, the machine would panic after a short while
with 0xfeedbeef in the normal dma buffer registers. A restart
cycle is probably hapenning without an interrupt or something.
minor of libc and the major of libutil). For little-endian architectures
merge the bnswap() assembly versions with nto* and hton* using symbols
aliasing. Use symbol renaming for the bswap function in this case to avoid
namespace pollution.
Declare bswap* in machine/bswap.h, not machine/endian.h. For little-endian
machines, common code for inline macros go in machine/byte_swap.h
Sync libkern with libc.
Adjust #include in kernel sources for machine/bswap.h.
Unfortunately, to get dma to work, we have to align past the end of
the buffer, which smashes other variables on the stack.
match function still disabled, as it isn't ready yet.
of DD_NEXT for regular dma transfers, and not just ethernet transmit.
Keep track of dma read/write direction and set it each time we start or
restart dma. This allows scsi to work, and doesn't appear to hinder ethernet.
- returned EOPNOTSUPP rather than -1.
- no check for negative offset.
many of these fix potential security problems in these drivers.
XXX XXX XXX
the d_mmap cdev routine should be changed to have a prototype like:
paddr_t (*d_mmap) __P((dev_t, off_t, int));
by someone!
- cpu_set_kpc() now takes void *arg third argument, passed to the
entry point.
- cpu_fork() allows parent to be non-curproc iff parent is proc0.
When forking non-curproc, assume its state has already been saved.
- Adjust various pieces of machine-dependent code to account of all of this.
-sys/lib/libkern builds as library per default (as it was documented all
the time)
-ports able to LKM set "KERN_AS=obj" explicitely in their Makefiles
(for now; should depend on actual "option LKM" or -better- functions
included for LKM use should be pulled in by a stub)
-always link libcompat before libkern - libkern stuff can be referred to
by libcompat, but not the other way
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.
which require special handling, e.g. sigreturn on m68k.
This differs from the old sigreturn trap in that we require the syscall
number to be in register d0, just like the regular syscall entry point.
This will allow sigreturn to be versioned in the future without the need
to allocate another trap vector.
u-area in machine-dependent code. Instead, call exit2() to schedule
the reaper to free them for us, once it is safe to do so (i.e. we are
no longer running on the dead proc's vmspace and stack).
