Commit Graph

1663 Commits

Author SHA1 Message Date
mycroft 9d0dcbced4 Prefault the u-area pages at swapin time. 1999-03-30 21:01:42 +00:00
mycroft dcdd0d6124 Map the system page with access_type=READ. 1999-03-30 20:59:52 +00:00
mycroft a3346d4ca8 Page 0 should never get R/M emulation. 1999-03-30 15:13:42 +00:00
mycroft 8cab76a0ec Make the kernel silent by default when running crashme. 1999-03-30 10:10:57 +00:00
mycroft f317021d2c Some of the DIAGNOSTIC checks were too strict and could be tripped by crashme. 1999-03-30 10:10:22 +00:00
mrg c17a4018cb pull in new (?) BUS_SPACE_BARRIER_{READ,WRITE} definitions. 1999-03-29 12:42:51 +00:00
mycroft 665b05efd4 GC physical_memoryblock. 1999-03-29 10:02:19 +00:00
mycroft e89e34c04a Nuke the contents of pmap_pageable(). It accomplishes nothing on this port,
and it's potentially a serious bug.
1999-03-29 07:15:06 +00:00
mycroft 460d63062a Do the previously slightly differently; we want to count by pages anyway.
Also fixes a problem with not flushing TLB entries in vmapbuf().
1999-03-29 06:24:31 +00:00
mycroft 12f5abbcee A few things:
* Count page table pages in the RSS.
* Rather than patching it up, panic if access_type has bits not in prot, as
  this should now be impossible.
* Add page table reference counting, but currently disabled as it still has
  some issues.
1999-03-28 22:01:06 +00:00
sommerfe ee7bbe5b2f Fix bug in previous change which resulted in kernel VA space leak.
We need an accurate len after we clear the PTE's in vunmapbuf.
1999-03-28 20:21:51 +00:00
mycroft ea8db6d88d Fix the vunmapbuf() problem, by zeroing out the PTEs before freeing the VA
range.  This is reasonable given that we inserted the PTEs manually in the
first place.
Also try to use specific cache/TLB purges.
1999-03-28 06:35:38 +00:00
nathanw 95b075be1e Remove commented-out uaudio declaration;
we don't actually have support for it.
1999-03-27 23:27:57 +00:00
mycroft 237c992085 Fix a possible glitch (not mine) in the wired page accounting. 1999-03-27 14:13:42 +00:00
mycroft 15217d40e3 More DIAGNOSTIC checks... 1999-03-27 11:45:07 +00:00
mycroft a1784400cd GC PT_Us, add more DIAGNOSTIC checks. 1999-03-27 09:41:03 +00:00
explorer 02db718d0f enable pseudo-device rnd on GENERIC-type, and remove EXPERIMENTAL from rest. 1999-03-27 07:11:35 +00:00
mycroft 93768384ee Fix a bug in pmap_collect_pv(), even though nobody currently uses it:
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.
1999-03-27 05:57:02 +00:00
mycroft 32295b2dac Further cleanup:
pmap_find_pv(), pmap_clean_page() and pmap_remove_all() are only called on
managed pages, after VM initialization.  Panic if this invariant is violated.
Also, panic if we try to enter a PT page through pmap_enter(), rather than
silently patching it up.
pmap_initialized is now #ifdef DIAGNOSTIC.
1999-03-27 05:12:21 +00:00
mycroft 9c6b797796 Changes for modified pmap_enter() API:
* 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.
1999-03-26 23:41:25 +00:00
mycroft e84fe91c51 Take advantage of the new `access_type' for pmap_enter(), and always do R/M
emulation of managed pages.  This required the following `interesting' changes:
* File system buffers must be entered with an access type of
  VM_PROT_READ|VM_PROT_WRITE, so that the pages will be accessible immediately.
  Otherwise we would have to teach pagemove() to update the R/M information.
  Since they're never eligible for paging, the latter is overkill.
* We must insure that pages allocated before the pmap is completely set up
  (that is, pages allocated early by the VM system) are not eligible for R/M
  emulation, since the memory needed for this isn't available.  We do this by
  allocating the pmap's internal memory with uvm_pageboot_alloc().  This also
  fixes an absolutely horrible hack where the pmap only worked because page 0
  happened to be mapped.
  to be mapped.
Also:
* Push the wired page counting into the p->v list maintenance functions.  This
  avoids code duplication, and fixes some cases where we were confused about
  which pages to do it with.
* Fix lots of problems associated with pmap_nightmare() (and rename it to
  pmap_vac_me_harder()).
* Since the early pages are no longer considered `managed', just make
  pmap_*_pv() panic if !pmap_initialized.
1999-03-26 22:00:24 +00:00
hubertf b97ebf9749 Sync some options from GENERIC 1999-03-26 02:48:18 +00:00
hubertf b3acf3a4e4 s/internally/internal/ 1999-03-26 02:46:14 +00:00
thorpej 587eda7c3f Use ETHER_*_LEN constants from <net/if_ether.h> instead of defining them
ourselves.
1999-03-25 23:11:51 +00:00
tron 732f052e13 Don't include "opt_uvm.h" any more. 1999-03-24 12:41:27 +00:00
tron 4cdff58b76 Don't include "opt_uvm.h" any more. 1999-03-24 12:16:50 +00:00
tron 75377e1815 Don't include "opt_uvm.h" any more. 1999-03-24 11:35:30 +00:00
mrg ca5f9685bb clean up kernel/config files files for machVM lossage. 1999-03-24 06:06:09 +00:00
mrg d2397ac5f7 completely remove Mach VM support. all that is left is the all the
header files as UVM still uses (most of) these.
1999-03-24 05:50:49 +00:00
mycroft 006ee23e99 Fix another problem is the modified bit emulation that caused double faults.
We need to set the L2 type preemptively, as it may still be L2_INVAL if the
page had not been previously accessed.
1999-03-24 02:45:27 +00:00
mycroft 11d7efb7ba Forgot a return instruction in the error case. 1999-03-24 01:51:56 +00:00
drochner 76fa1751fe Add a macro to check for sufficient pointer alignment in bus.h context,
"BUS_SPACE_ALIGNED_POINTER()".
Equal to the param.h "ALIGNED_POINTER()" normally, but obeys additional
requirements of the bus_space_xxx_n() macros. (BUS_SPACE_DEBUG)
1999-03-23 21:29:03 +00:00
mycroft 1bb64b0636 __SWAP_BROKEN is no longer useful here. 1999-03-23 19:16:03 +00:00
mycroft 6bb727e736 Nuke some more code that was fetching PT_M and PT_H from pv_flags. 1999-03-23 18:39:38 +00:00
mycroft 5f63cab877 The page and section fault handling paths are essentially identical, so
combine them.  This also fixes a performance issue, in that we always
faulted pages twice in order to write them.
1999-03-23 18:02:02 +00:00
mycroft 79e56d3237 The ARM has the same copy-on-write bug as the i386 -- the write enable bit in
PTEs is ignored when in kernel mode.  Hack around this just like we do on the
i386, by adding a prepass to copyout() to check for write permission on the
destination pages.
1999-03-23 17:14:34 +00:00
mycroft 8f3d31a32b More cleanup...
* Don't bother pulling PT_M and PT_H bits from pv_flags; they can't ever be
  set there!
* Actually make pmap_clear_reference() do something useful.
* Also set the referenced bit (PT_H) when emulating a write fault.
1999-03-23 13:52:48 +00:00
mycroft da99fe308e Repeat after me:
If we're doing modified bit emulation, we must revoke write permission in
pmap_clear_modify().  This is non-negotiable.  I will revoke write permission
in pmap_clear_modify(), or suffer the wrath of a thousand bricks.
1999-03-23 13:27:48 +00:00
mycroft 3c8512dbb5 We could not possibly have been updating the pv attributed correctly in
pmap_enter(), so... make the obvious change.
1999-03-23 12:30:45 +00:00
mark f88d8c6a35 In the light of PRs 7122 & 7123 temporarily define __SWAP_BROKEN until
the problem these bug reports described is fixed and they are closed.
1999-03-22 10:24:04 +00:00
mark 34a92d6f82 Fix addressing problem for the aux status register. Whilst reads of this
read appear to be aliased in the address space, writes are not thus the
correct address is important in order to reset drives.
1999-03-22 10:14:12 +00:00
thorpej a77ccfe460 Garbage-collect. 1999-03-20 01:40:25 +00:00
cgd d324c9f2f6 pull isa_machdep.h in from machine/ instead of from $(MACHINE)/isa/ 1999-03-19 05:13:16 +00:00
cgd b47008cec9 Moved to arch/arm32/include/isa_machdep.h,v 1999-03-19 05:01:52 +00:00
cgd 04216fffa4 Moved from arch/arm32/isa/isa_machdep.h,v 1999-03-19 05:01:51 +00:00
cgd d00106ee06 if you pull in isavar.h, you don't need isa_machdep.h. 1999-03-19 04:58:45 +00:00
cgd f753c9860f Moved to arch/arm32/include/pci_machdep.h,v 1999-03-19 03:34:22 +00:00
cgd 4b75fd0ceb Moved from arch/arm32/pci/pci_machdep.h,v 1999-03-19 03:34:21 +00:00
cgd 6ce0de2b8e Moved to arch/arm32/include/ofisa_machdep.h,v 1999-03-19 03:29:50 +00:00
cgd 9c73a61eca Moved from arch/arm32/ofw/ofisa_machdep.h,v 1999-03-19 03:29:49 +00:00
cgd b95dc1d904 Moved to arch/arm32/include/isapnp_machdep.h,v 1999-03-19 03:14:17 +00:00
cgd 126a9a68b7 Moved from arch/arm32/isa/isapnp_machdep.h,v 1999-03-19 03:14:16 +00:00
chs ab7269f62a if uvm_fault() fails with KERN_RESOURCE_SHORTAGE, send a SIGKILL
and print a message about it.  this will be used to recover from
out-of-swap conditions.
1999-03-18 04:56:01 +00:00
sommerfe bc533621ed defopt MINIROOTSIZE 1999-03-17 18:59:21 +00:00
mark 2ae8daf4dc Junk a load of register keywords.
Add missing type in a declaration of iobase.
1999-03-16 10:55:42 +00:00
mark 1b8eea513f _intrnames is only used in irq_release() if IRQ_STATS is defined so add an
appropriate guard.
1999-03-16 10:53:50 +00:00
perry 2bbefd75ec remove ovbcopy reference 1999-03-12 23:05:42 +00:00
mark 303ca6c449 Make pmap_extract() understand L2 large page PTEs. Normally small page
PTEs are used but during bootstrap the kernel may be mapped with large page
PTE to save TLB entries.
1999-03-10 20:54:17 +00:00
mycroft 8a90f14119 Brace ourselves for egcs... 1999-03-06 01:29:53 +00:00
lukem 6bf077c969 support BUFCACHE (a la the work i did for the i386, etc)
XXX: this worked a while ago when i last tested, but i haven't
compiled it recently. the code looks sane, however
1999-03-05 03:26:36 +00:00
mark 98c62c02c3 Since all the calls to the postmortem debug code have been removed by
various folks, the actual code can be removed as well.
1999-03-01 10:01:52 +00:00
mark d4a3aa46b4 Add support for tracing other processes with trace /t. 1999-03-01 05:48:31 +00:00
mark 3aca0028f1 Fix else ambiguity. 1999-02-28 10:01:08 +00:00
mark 8b41c31868 Fix compiler warning. 1999-02-28 09:59:33 +00:00
mark b8ca758af8 Implement pmap_procwr() the arm32 part of fix for pr 6152. 1999-02-28 01:49:25 +00:00
mark d6f8b575ea Add PCI-PCI bridge device.
Up maxusers to 32.
1999-02-27 11:19:25 +00:00
mark bbd7276407 Add PCI-PCI bridge devices. 1999-02-27 11:18:27 +00:00
mark 54e053627e Implement bus_space_{read,write}_region_2().
Make bus_space_copy_region_2() work correctly for overlapping regions.
1999-02-27 11:14:27 +00:00
mark 77d06b9612 Remove a debug hack. 1999-02-27 11:12:44 +00:00
scottr d32ed292af defopt BUFCACHE and BUFPAGES. 1999-02-27 06:39:34 +00:00
is d93a991a48 Make cpu_model big enough to hold the DNARD cpu's identification, and a bit
more.
1999-02-26 12:05:49 +00:00
mycroft 45e082f472 Fix thinko. 1999-02-19 22:32:21 +00:00
mycroft 7d0a7c6abf Do the pcb_onfault test earlier, so that the DIAGNOSTIC code doesn't trip. 1999-02-19 22:15:09 +00:00
hubertf 100778d54a RCS ID police 1999-02-15 04:25:28 +00:00
kleink f69591d12d Use of casts as lvalues is a GNU C extension; rearrange slightly. 1999-02-10 17:03:26 +00:00
jonathan c9d036d796 defopt MEMORY_DISK_{HOOKS,SERVER,IS_ROOT}. 1999-02-07 09:34:58 +00:00
mark 5ef478be57 Fix some egcs compiler warnings (from Patrick Welche).
Closes PR6925.
1999-02-02 04:02:26 +00:00
tron 4fe3a6480e Fix bug exposed by new "mkdep" program:
"${SFILES}" is empty for a "SHARK" kernel so that "cc" will fail with
"No input files" during "mkdep".
1999-01-29 22:20:49 +00:00
matthias b88b2d4f9b MBR_PTYP_* -> MBR_PTYPE_* to make this compile again. 1999-01-29 21:20:22 +00:00
thorpej 7ad46b1420 Goodbye disklabel_mbr.h. 1999-01-27 22:08:09 +00:00
thorpej 41c5275965 Use the MI disklabel_mbr.h. 1999-01-27 20:43:26 +00:00
thorpej 8851ed323e Fix a bug that has annoyed me for a long time. In pmap_enter(), if
there are no pages available immediately to use as a PT page, don't
just roll over and die.  Only do that if we're the kernel pmap (shouldn't
happen very ofter [ever?!], and I'm not certain we're guaranteed valid
thread context when operating on the kernel pmap).  For user pmaps, wait
for the pagedaemon to wake us up when more free pages are available.
1999-01-26 09:03:31 +00:00
sommerfe d3eec1e6c2 nit: Fix a missed pms->opms rename (noticed by Matthias). 1999-01-24 18:58:12 +00:00
mycroft 897a8a7f29 Mark [hn]to[nh][ls]() with __const__, so they are subject to CSE. 1999-01-24 12:55:01 +00:00
sommerfe 83940dae72 Finish the rest of the pms->opms rename for non-wscons PS/2 mice. 1999-01-23 22:18:41 +00:00
drochner 6fe2522d57 second round of mouse renaming: psm->pms 1999-01-23 16:10:47 +00:00
mycroft 7497ccd364 Clean up. 1999-01-22 14:19:54 +00:00
mycroft ec253b6315 Use __builtin_next_arg(). Fixed PR 6862. 1999-01-22 14:14:32 +00:00
mark f6f7ac980d Fix bug in the initialization of the interrupt masks for shared
interrupts as described in PR6421. Fix from Richard Earnshaw.
1999-01-22 06:59:26 +00:00
mycroft 1c607f81a8 Fix editing error. 1999-01-21 08:20:25 +00:00
mycroft cefdcd4840 More DDB cleanup. 1999-01-20 13:56:35 +00:00
mark 8ee86da26a Clean up the last remains of non-MNN the code. 1999-01-17 06:58:16 +00:00
chuck 4e484e0e02 remove dead non-MNN code (one block left) 1999-01-16 21:03:48 +00:00
chuck 66226eb6ff remove non-MNN code 1999-01-16 20:11:03 +00:00
thorpej 8922647c58 Some minor, mostly costmetic, changes to CPPFLAGS/CFLAGS. 1999-01-15 23:37:05 +00:00
bouyer dc306354b0 Move the bswap functions from libutil to libc (this bups the
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.
1999-01-15 13:31:15 +00:00
mark a82d8b0a33 Fix void * arithmetic. 1999-01-09 22:56:06 +00:00
thorpej e598335d1c Garbage-collect `mbutl'. 1999-01-09 22:10:12 +00:00
mark 444401c526 Fix `void *' arithmetic. 1999-01-09 21:18:54 +00:00
mark e6361430fc Fix `void *' arithmetic in bcopy() calls. 1999-01-09 21:10:50 +00:00