]> Zhao Yanbai Git Server - minix.git/log
minix.git
18 years ago*** empty log message ***
Ben Gras [Wed, 8 Mar 2006 16:45:55 +0000 (16:45 +0000)]
*** empty log message ***

18 years ago*** empty log message ***
Ben Gras [Wed, 8 Mar 2006 16:34:36 +0000 (16:34 +0000)]
*** empty log message ***

18 years agoupdate needs more stack in order to handle signals
Ben Gras [Wed, 8 Mar 2006 15:42:31 +0000 (15:42 +0000)]
update needs more stack in order to handle signals

18 years agoDon't blindly skip 0-group processes in signals
Ben Gras [Wed, 8 Mar 2006 14:46:42 +0000 (14:46 +0000)]
Don't blindly skip 0-group processes in signals

18 years agoendpoint fixes for RS
Ben Gras [Wed, 8 Mar 2006 14:38:35 +0000 (14:38 +0000)]
endpoint fixes for RS

18 years agoRS formatting fixes
Ben Gras [Wed, 8 Mar 2006 14:37:26 +0000 (14:37 +0000)]
RS formatting fixes

18 years agoEndpoint-enabled fix
Ben Gras [Wed, 8 Mar 2006 13:10:35 +0000 (13:10 +0000)]
Endpoint-enabled fix

18 years agoFixed typo.
Philip Homburg [Wed, 8 Mar 2006 12:30:43 +0000 (12:30 +0000)]
Fixed typo.

18 years agorc: don't start floppy or at driver
Ben Gras [Wed, 8 Mar 2006 12:21:40 +0000 (12:21 +0000)]
rc: don't start floppy or at driver

passwd: add games for nethack

18 years agoAdded some disabled debug code.
Philip Homburg [Wed, 8 Mar 2006 12:06:33 +0000 (12:06 +0000)]
Added some disabled debug code.

18 years agoDo not confuse proc with proc_addr if you want a bug free kernel.
Philip Homburg [Wed, 8 Mar 2006 11:59:59 +0000 (11:59 +0000)]
Do not confuse proc with proc_addr if you want a bug free kernel.

18 years agoTest receive() return code
Ben Gras [Tue, 7 Mar 2006 15:45:14 +0000 (15:45 +0000)]
Test receive() return code

18 years agoTightened generation size - don't count process numbers more than once
Ben Gras [Tue, 7 Mar 2006 15:44:34 +0000 (15:44 +0000)]
Tightened generation size - don't count process numbers more than once

18 years agoImproved test constant
Ben Gras [Tue, 7 Mar 2006 15:44:00 +0000 (15:44 +0000)]
Improved test constant

18 years agoTake out arbitrary constant
Ben Gras [Tue, 7 Mar 2006 15:38:43 +0000 (15:38 +0000)]
Take out arbitrary constant

18 years agoFixed wrong fix
Ben Gras [Tue, 7 Mar 2006 14:58:18 +0000 (14:58 +0000)]
Fixed wrong fix

18 years agoInclude boot options
Ben Gras [Tue, 7 Mar 2006 14:32:41 +0000 (14:32 +0000)]
Include boot options

18 years agoFixed masks on I/O ports.
Philip Homburg [Tue, 7 Mar 2006 14:16:13 +0000 (14:16 +0000)]
Fixed masks on I/O ports.

18 years agoDisabled pci_intel_ctrl.
Philip Homburg [Tue, 7 Mar 2006 14:14:53 +0000 (14:14 +0000)]
Disabled pci_intel_ctrl.

18 years agoAdded rawspeed.
Philip Homburg [Tue, 7 Mar 2006 14:12:51 +0000 (14:12 +0000)]
Added rawspeed.

18 years agoLess stack for ds and rs
Ben Gras [Tue, 7 Mar 2006 13:22:36 +0000 (13:22 +0000)]
Less stack for ds and rs

18 years agoAlign on 16 byte boundary, not 64k
Ben Gras [Tue, 7 Mar 2006 12:38:51 +0000 (12:38 +0000)]
Align on 16 byte boundary, not 64k

18 years agoDo alignment after 64k-dma-boundary adjustment
Ben Gras [Mon, 6 Mar 2006 16:46:12 +0000 (16:46 +0000)]
Do alignment after 64k-dma-boundary adjustment

18 years agoPrint PCI capability types.
Philip Homburg [Mon, 6 Mar 2006 15:19:51 +0000 (15:19 +0000)]
Print PCI capability types.

18 years agoA bit of leftover cleanup work.
Philip Homburg [Mon, 6 Mar 2006 10:42:51 +0000 (10:42 +0000)]
A bit of leftover cleanup work.

18 years agoComment fixes mailed to bugs@
Ben Gras [Sat, 4 Mar 2006 22:51:52 +0000 (22:51 +0000)]
Comment fixes mailed to bugs@

18 years agoGeneration number printed too
Ben Gras [Fri, 3 Mar 2006 16:34:37 +0000 (16:34 +0000)]
Generation number printed too

18 years agoBuild memory driver last.
Philip Homburg [Fri, 3 Mar 2006 15:38:27 +0000 (15:38 +0000)]
Build memory driver last.

18 years agoSupport for DMA.
Philip Homburg [Fri, 3 Mar 2006 15:37:51 +0000 (15:37 +0000)]
Support for DMA.

18 years agoAlways do sanity check now. (Possible now that we have lba48.)
Ben Gras [Fri, 3 Mar 2006 15:34:24 +0000 (15:34 +0000)]
Always do sanity check now. (Possible now that we have lba48.)

18 years agoAdded 'xxl' binsizes size for x.
Ben Gras [Fri, 3 Mar 2006 12:44:22 +0000 (12:44 +0000)]
Added 'xxl' binsizes size for x.
Leave sh and make at normal size for 'big'.
Call binsizes normal after creating release.

18 years agoEndpoint field name
Ben Gras [Fri, 3 Mar 2006 12:13:37 +0000 (12:13 +0000)]
Endpoint field name

18 years agoFix of endpointless slot message in FS
Ben Gras [Fri, 3 Mar 2006 10:25:28 +0000 (10:25 +0000)]
Fix of endpointless slot message in FS

Removed some leftover '#if 0'ed debug code in PM

18 years ago*** empty log message ***
Ben Gras [Fri, 3 Mar 2006 10:22:35 +0000 (10:22 +0000)]
*** empty log message ***

18 years agoendpoint migration for drivers.
Ben Gras [Fri, 3 Mar 2006 10:21:45 +0000 (10:21 +0000)]
endpoint migration for drivers.

mostly renaming message field names to the new names.

tty stored process numbers in chars in tty and pty structs - now ints.

18 years agoendpoint-aware conversion of servers.
Ben Gras [Fri, 3 Mar 2006 10:20:58 +0000 (10:20 +0000)]
endpoint-aware conversion of servers.

'who', indicating caller number in pm and fs and some other servers, has
been removed in favour of 'who_e' (endpoint) and 'who_p' (proc nr.).

In both PM and FS, isokendpt() convert endpoints to process slot
numbers, returning OK if it was a valid and consistent endpoint number.
okendpt() does the same but panic()s if it doesn't succeed. (In PM,
this is pm_isok..)

pm and fs keep their own records of process endpoints in their proc tables,
which are needed to make kernel calls about those processes.

message field names have changed.

fs drivers are endpoints.

fs now doesn't try to get out of driver deadlock, as the protocol isn't
supposed to let that happen any more. (A warning is printed if ELOCKED
is detected though.)

fproc[].fp_task (indicating which driver the process is suspended on)
became an int.

PM and FS now get endpoint numbers of initial boot processes from the
kernel. These happen to be the same as the old proc numbers, to let
user processes reach them with the old numbers, but FS and PM don't know
that. All new processes after INIT, even after the generation number
wraps around, get endpoint numbers with generation 1 and higher, so
the first instances of the boot processes are the only processes ever
to have endpoint numbers in the old proc number range.

More return code checks of sys_* functions have been added.

IS has become endpoint-aware. Ditched the 'text' and 'data' fields
in the kernel dump (which show locations, not sizes, so aren't terribly
useful) in favour of the endpoint number. Proc number is still visible.

Some other dumps (e.g. dmap, rs) show endpoint numbers now too which got
the formatting changed.

PM reading segments using rw_seg() has changed - it uses other fields
in the message now instead of encoding the segment and process number and
fd in the fd field. For that it uses _read_pm() and _write_pm() which to
_taskcall()s directly in pm/misc.c.

PM now sys_exit()s itself on panic(), instead of sys_abort().

RS also talks in endpoints instead of process numbers.

18 years ago'proc number' is process slot, 'endpoint' are generation-aware process
Ben Gras [Fri, 3 Mar 2006 10:00:02 +0000 (10:00 +0000)]
'proc number' is process slot, 'endpoint' are generation-aware process
instance numbers, encoded and decoded using macros in <minix/endpoint.h>.

proc number -> endpoint migration
  . proc_nr in the interrupt hook is now an endpoint, proc_nr_e.
  . m_source for messages and notifies is now an endpoint, instead of
    proc number.
  . isokendpt() converts an endpoint to a process number, returns
    success (but fails if the process number is out of range, the
    process slot is not a living process, or the given endpoint
    number does not match the endpoint number in the process slot,
    indicating an old process).
  . okendpt() is the same as isokendpt(), but panic()s if the conversion
    fails. This is mainly used for decoding message.m_source endpoints,
    and other endpoint numbers in kernel data structures, which should
    always be correct.
  . if DEBUG_ENABLE_IPC_WARNINGS is enabled, isokendpt() and okendpt()
    get passed the __FILE__ and __LINE__ of the calling lines, and
    print messages about what is wrong with the endpoint number
    (out of range proc, empty proc, or inconsistent endpoint number),
    with the caller, making finding where the conversion failed easy
    without having to include code for every call to print where things
    went wrong. Sometimes this is harmless (wrong arg to a kernel call),
    sometimes it's a fatal internal inconsistency (bogus m_source).
  . some process table fields have been appended an _e to indicate it's
    become and endpoint.
  . process endpoint is stored in p_endpoint, without generation number.
    it turns out the kernel never needs the generation number, except
    when fork()ing, so it's decoded then.
  . kernel calls all take endpoints as arguments, not proc numbers.
    the one exception is sys_fork(), which needs to know in which slot
    to put the child.

18 years ago. rename message fields to endpoint variants
Ben Gras [Fri, 3 Mar 2006 09:44:55 +0000 (09:44 +0000)]
. rename message fields to endpoint variants
. remove or optionalify some pci debugging stuff

18 years agoChanges to support endpoints.
Ben Gras [Fri, 3 Mar 2006 09:43:06 +0000 (09:43 +0000)]
Changes to support endpoints.
  . rename message field name macros from *PROC* to *ENDPT*,
    both to reflect the new meaning and to use it to hunt down where
    they are used
  . _PM_SEG_FLAG in fd replaces funny segment loading construction by
    PM in FS
  . _MAX_MAGIC_PROC is the highest used proc number, including magic
    constants such as (currently) ANY, NONE and SELF, used by the
    endpoint macros to make sure they are encoded properly and no valid
    endpoint number ever encodes to it.
  . rename proc_nr in address copy struct to proc_nr_e (endpoint)

18 years agoTest fixes
Ben Gras [Wed, 1 Mar 2006 15:28:07 +0000 (15:28 +0000)]
Test fixes

Added test41, for endpoint macros

18 years agoDon't include cdprobe any more
Ben Gras [Fri, 24 Feb 2006 12:56:06 +0000 (12:56 +0000)]
Don't include cdprobe any more

18 years agoLBA48 support.
Philip Homburg [Fri, 24 Feb 2006 12:55:47 +0000 (12:55 +0000)]
LBA48 support.

18 years agoChange to some debug code.
Philip Homburg [Fri, 24 Feb 2006 12:55:00 +0000 (12:55 +0000)]
Change to some debug code.

18 years agoAllow bigger sizes.
Philip Homburg [Fri, 24 Feb 2006 12:53:54 +0000 (12:53 +0000)]
Allow bigger sizes.

18 years agoLance needs 4k due to 4k clicksize
Ben Gras [Wed, 22 Feb 2006 13:45:29 +0000 (13:45 +0000)]
Lance needs 4k due to 4k clicksize

18 years agoDon't complain, urlget
Ben Gras [Tue, 21 Feb 2006 16:06:28 +0000 (16:06 +0000)]
Don't complain, urlget

18 years agoPeter Boonstoppel's recwave update.
Ben Gras [Tue, 21 Feb 2006 16:04:18 +0000 (16:04 +0000)]
Peter Boonstoppel's recwave update.

18 years agoPeter Boonstoppel's audio changes.
Ben Gras [Tue, 21 Feb 2006 15:59:33 +0000 (15:59 +0000)]
Peter Boonstoppel's audio changes.

18 years agoInitial endpoint macros.
Ben Gras [Tue, 21 Feb 2006 15:26:12 +0000 (15:26 +0000)]
Initial endpoint macros.

18 years ago*** empty log message ***
Ben Gras [Mon, 20 Feb 2006 16:00:24 +0000 (16:00 +0000)]
*** empty log message ***

18 years agoIt's not always initialization
Ben Gras [Mon, 20 Feb 2006 15:11:41 +0000 (15:11 +0000)]
It's not always initialization

18 years agoFixed bug in canceling TCP ioctls before INET got an IP address.
Philip Homburg [Mon, 20 Feb 2006 14:35:01 +0000 (14:35 +0000)]
Fixed bug in canceling TCP ioctls before INET got an IP address.

18 years agoDon't copy memory to rescue
Ben Gras [Fri, 17 Feb 2006 16:46:08 +0000 (16:46 +0000)]
Don't copy memory to rescue

local package initialization

18 years agoMake a /usr/local/etc/rc.d
Ben Gras [Fri, 17 Feb 2006 15:14:46 +0000 (15:14 +0000)]
Make a /usr/local/etc/rc.d

18 years agoInstall memory in /usr/sbin
Ben Gras [Fri, 17 Feb 2006 15:03:38 +0000 (15:03 +0000)]
Install memory in /usr/sbin

18 years agoSmaller device
Ben Gras [Fri, 17 Feb 2006 14:43:00 +0000 (14:43 +0000)]
Smaller device

18 years agoConditional compilation for SIGIOT and SIGEMT.
Philip Homburg [Fri, 17 Feb 2006 14:12:47 +0000 (14:12 +0000)]
Conditional compilation for SIGIOT and SIGEMT.

18 years agoReintroduced SIGEMT.
Philip Homburg [Fri, 17 Feb 2006 14:12:12 +0000 (14:12 +0000)]
Reintroduced SIGEMT.

18 years agoRefix
Ben Gras [Fri, 17 Feb 2006 14:05:46 +0000 (14:05 +0000)]
Refix

18 years agoimproved tcp_acceptto.
Philip Homburg [Fri, 17 Feb 2006 13:47:31 +0000 (13:47 +0000)]
improved tcp_acceptto.

18 years agoSocket options SO_RCVBUF, SO_SNDBUF, and TCP_NODELAY.
Philip Homburg [Fri, 17 Feb 2006 13:45:41 +0000 (13:45 +0000)]
Socket options SO_RCVBUF, SO_SNDBUF, and TCP_NODELAY.

18 years agoMore error texts.
Philip Homburg [Fri, 17 Feb 2006 13:44:48 +0000 (13:44 +0000)]
More error texts.

18 years agoNew errors and defines for socket options.
Philip Homburg [Fri, 17 Feb 2006 13:43:50 +0000 (13:43 +0000)]
New errors and defines for socket options.

18 years agoSIGBUS needs its own number. Removed SIGEMT. Made SIGIOT conditional on
Philip Homburg [Fri, 17 Feb 2006 13:43:07 +0000 (13:43 +0000)]
SIGBUS needs its own number. Removed SIGEMT. Made SIGIOT conditional on
_MINIX.

18 years agodon't make clean first
Ben Gras [Fri, 17 Feb 2006 13:27:48 +0000 (13:27 +0000)]
don't make clean first

18 years agoThrow elvis out of the base system
Ben Gras [Fri, 17 Feb 2006 13:27:01 +0000 (13:27 +0000)]
Throw elvis out of the base system

18 years agomemory moved
Ben Gras [Fri, 17 Feb 2006 13:19:10 +0000 (13:19 +0000)]
memory moved

18 years agoMake 'make install' work
Ben Gras [Fri, 17 Feb 2006 13:12:07 +0000 (13:12 +0000)]
Make 'make install' work

18 years agoLet saving random data work again
Ben Gras [Fri, 17 Feb 2006 13:11:20 +0000 (13:11 +0000)]
Let saving random data work again

18 years ago*** empty log message ***
Ben Gras [Fri, 17 Feb 2006 13:10:37 +0000 (13:10 +0000)]
*** empty log message ***

18 years agoKick nvi out of base
Ben Gras [Fri, 17 Feb 2006 12:36:53 +0000 (12:36 +0000)]
Kick nvi out of base

18 years agoMake more
Ben Gras [Fri, 17 Feb 2006 12:34:18 +0000 (12:34 +0000)]
Make more

18 years agoAdded some of my root devices
Ben Gras [Fri, 17 Feb 2006 11:58:02 +0000 (11:58 +0000)]
Added some of my root devices

18 years agogetmachine info first, then init tty.
Ben Gras [Fri, 17 Feb 2006 10:23:16 +0000 (10:23 +0000)]
getmachine info first, then init tty.
Problem report and fix given by:
Prof: Alejandro T. Bello Ruiz <abello@macareo.pucp.edu.pe>

18 years agobigmake.inc is obsolete
Ben Gras [Thu, 16 Feb 2006 20:56:34 +0000 (20:56 +0000)]
bigmake.inc is obsolete

18 years ago*** empty log message ***
Ben Gras [Thu, 16 Feb 2006 15:44:52 +0000 (15:44 +0000)]
*** empty log message ***

18 years agoFixes for packaging system that will read .postinstall script and run it
Ben Gras [Thu, 16 Feb 2006 15:43:37 +0000 (15:43 +0000)]
Fixes for packaging system that will read .postinstall script and run it
if exists.

packit installs a package.

18 years agosshd user
Ben Gras [Thu, 16 Feb 2006 15:38:13 +0000 (15:38 +0000)]
sshd user

18 years agoFixed two memory leaks.
Philip Homburg [Thu, 16 Feb 2006 14:13:36 +0000 (14:13 +0000)]
Fixed two memory leaks.

18 years agoReally disable debug
Ben Gras [Wed, 15 Feb 2006 14:22:40 +0000 (14:22 +0000)]
Really disable debug

18 years agoMoved PCI driver to the ramdisk.
Philip Homburg [Wed, 15 Feb 2006 14:21:56 +0000 (14:21 +0000)]
Moved PCI driver to the ramdisk.

18 years agoTurn off debug in getsockname.c
Ben Gras [Wed, 15 Feb 2006 14:16:06 +0000 (14:16 +0000)]
Turn off debug in getsockname.c

allow NULL for setgroups argument

18 years agoInitial root filesystem is now on a ramdisk that is part of the image.
Philip Homburg [Wed, 15 Feb 2006 11:18:21 +0000 (11:18 +0000)]
Initial root filesystem is now on a ramdisk that is part of the image.

18 years agoimage_small doesn't fit anymore. What do we do with that?
Philip Homburg [Wed, 15 Feb 2006 11:06:32 +0000 (11:06 +0000)]
image_small doesn't fit anymore. What do we do with that?

18 years ago/dev/ram is no longer hardcoded.
Philip Homburg [Wed, 15 Feb 2006 11:05:47 +0000 (11:05 +0000)]
/dev/ram is no longer hardcoded.

18 years agoAccept optional device parameter for cdfdboot.
Philip Homburg [Wed, 15 Feb 2006 11:03:28 +0000 (11:03 +0000)]
Accept optional device parameter for cdfdboot.

18 years agoMake initgroups() fail if user has any supplemental groups.
Ben Gras [Mon, 13 Feb 2006 15:06:01 +0000 (15:06 +0000)]
Make initgroups() fail if user has any supplemental groups.

18 years agoFake setgroups() and initgroups() implementation.
Ben Gras [Mon, 13 Feb 2006 15:00:49 +0000 (15:00 +0000)]
Fake setgroups() and initgroups() implementation.

18 years ago*** empty log message ***
Ben Gras [Mon, 13 Feb 2006 13:10:53 +0000 (13:10 +0000)]
*** empty log message ***

18 years agoAdded truncate/ftruncate manpage.
Ben Gras [Mon, 13 Feb 2006 12:53:50 +0000 (12:53 +0000)]
Added truncate/ftruncate manpage.

18 years ago. changed MAXOPEN's from statics to (OPEN_MAX-3)
Ben Gras [Mon, 13 Feb 2006 11:07:21 +0000 (11:07 +0000)]
. changed MAXOPEN's from statics to (OPEN_MAX-3)
  (test broke after increasing OPEN_MAX)
. changed s_nokill to s_nokill_stop
  (test broken as SIGSTOP isn't maskable)

18 years agoRevert to 30 OPEN_MAX (committed by mistake)
Ben Gras [Mon, 13 Feb 2006 10:32:05 +0000 (10:32 +0000)]
Revert to 30 OPEN_MAX (committed by mistake)

18 years agoDon't sync before exiting drivers.
Ben Gras [Mon, 13 Feb 2006 10:28:42 +0000 (10:28 +0000)]
Don't sync before exiting drivers.

Wipe inodes of pipes in truncate_inode(), to avoid the size remaining
at PIPE_SIZE causing pipes to break.

18 years agoadded reenter check to lock_dequeue() to avoid unlocking of interrupts
Ben Gras [Fri, 10 Feb 2006 16:53:51 +0000 (16:53 +0000)]
added reenter check to lock_dequeue() to avoid unlocking of interrupts
via cause_sig() during an exception.

moved lock check configuration to <minix/sys_config.h> instead of
kernel/config.h, because the 'relocking' field in kinfo depends on it.

other prettification: common locking macro, whether lock timing is on or
not.

18 years agoumount /usr first
Ben Gras [Mon, 6 Feb 2006 16:40:15 +0000 (16:40 +0000)]
umount /usr first

18 years agoRun postinstall script
Ben Gras [Mon, 6 Feb 2006 16:17:54 +0000 (16:17 +0000)]
Run postinstall script

18 years agoIgnore system signals.
Ben Gras [Mon, 6 Feb 2006 15:33:39 +0000 (15:33 +0000)]
Ignore system signals.

18 years agoMoved ipc debug out of conditional.
Ben Gras [Mon, 6 Feb 2006 15:16:14 +0000 (15:16 +0000)]
Moved ipc debug out of conditional.
Gave pci server sys_getinfo privilege.

18 years ago#ifdeffed out some more debug stuff
Ben Gras [Mon, 6 Feb 2006 15:03:51 +0000 (15:03 +0000)]
#ifdeffed out some more debug stuff

18 years ago#ifdeffed out some debug code
Ben Gras [Mon, 6 Feb 2006 15:02:19 +0000 (15:02 +0000)]
#ifdeffed out some debug code