Ben Gras [Mon, 21 Sep 2009 14:31:52 +0000 (14:31 +0000)]
Primary goal for these changes is:
- no longer have kernel have its own page table that is loaded
on every kernel entry (trap, interrupt, exception). the primary
purpose is to reduce the number of required reloads.
Result:
- kernel can only access memory of process that was running when
kernel was entered
- kernel must be mapped into every process page table, so traps to
kernel keep working
Problem:
- kernel must often access memory of arbitrary processes (e.g. send
arbitrary processes messages); this can't happen directly any more;
usually because that process' page table isn't loaded at all, sometimes
because that memory isn't mapped in at all, sometimes because it isn't
mapped in read-write.
So:
- kernel must be able to map in memory of any process, in its own
address space.
Implementation:
- VM and kernel share a range of memory in which addresses of
all page tables of all processes are available. This has two purposes:
. Kernel has to know what data to copy in order to map in a range
. Kernel has to know where to write the data in order to map it in
That last point is because kernel has to write in the currently loaded
page table.
- Processes and kernel are separated through segments; kernel segments
haven't changed.
- The kernel keeps the process whose page table is currently loaded
in 'ptproc.'
- If it wants to map in a range of memory, it writes the value of the
page directory entry for that range into the page directory entry
in the currently loaded map. There is a slot reserved for such
purposes. The kernel can then access this memory directly.
- In order to do this, its segment has been increased (and the
segments of processes start where it ends).
- In the pagefault handler, detect if the kernel is doing
'trappable' memory access (i.e. a pagefault isn't a fatal
error) and if so,
- set the saved instruction pointer to phys_copy_fault,
breaking out of phys_copy
- set the saved eax register to the address of the page
fault, both for sanity checking and for checking in
which of the two ranges that phys_copy was called
with the fault occured
- Some boot-time processes do not have their own page table,
and are mapped in with the kernel, and separated with
segments. The kernel detects this using HASPT. If such a
process has to be scheduled, any page table will work and
no page table switch is done.
Major changes in kernel are
- When accessing user processes memory, kernel no longer
explicitly checks before it does so if that memory is OK.
It simply makes the mapping (if necessary), tries to do the
operation, and traps the pagefault if that memory isn't present;
if that happens, the copy function returns EFAULT.
So all of the CHECKRANGE_OR_SUSPEND macros are gone.
- Kernel no longer has to copy/read and parse page tables.
- A message copying optimisation: when messages are copied, and
the recipient isn't mapped in, they are copied into a buffer
in the kernel. This is done in QueueMess. The next time
the recipient is scheduled, this message is copied into
its memory. This happens in schedcheck().
This eliminates the mapping/copying step for messages, and makes
it easier to deliver messages. This eliminates soft_notify.
- Kernel no longer creates a page table at all, so the vm_setbuf
and pagetable writing in memory.c is gone.
Minor changes in kernel are
- ipc_stats thrown out, wasn't used
- misc flags all renamed to MF_*
- NOREC_* macros to enter and leave functions that should not
be called recursively; just sanity checks really
- code to fully decode segment selectors and descriptors
to print on exceptions
- lots of vmassert()s added, only executed if DEBUG_VMASSERT is 1
Ben Gras [Mon, 21 Sep 2009 14:25:54 +0000 (14:25 +0000)]
- tty: only report unrecognized scancodes once; forget about
remembering the origin and cursor position as that feature didn't
really work properly anyway
- tty: map in video and font memory using a vm call, access it from C,
thereby eliminating pesky weird segment calls and assembly to access it,
and unbreaks loadfont (Roman Ignatov)
- bios_wini: fix bios_wini by allocating a <1MB buffers for it
- memory: preallocate ramdisk, makes it a bit faster (and doesn't
fail halfway if you allocate a huge one)
- floppy: use <1MB buffer
- ramdisk proto: because of the 2x1 page reservations, binaries
got a little fatter and didn't fit on the ramdisk any more.
increase it.
Ben Gras [Mon, 21 Sep 2009 14:24:29 +0000 (14:24 +0000)]
- added 'datasizes' script that shows you the size allocated
for each symbol, usually answering those "why is does my binary have
such a lot of BSS" questions.
- stop binpackage looking in /var/spool for package files.
- let makewhatis recognize .Sh as heading name
- setup, fsck, df: allow >4kB block sizes painlessly
- mkfs: new #-of-inodes heuristic that depends on kb, not
on blocks; i've run out of inodes on my /usr
- asmconv: don't silently truncate .aligns to 16 bytes
- ipc* commands for shared memory support
Ben Gras [Mon, 21 Sep 2009 14:23:47 +0000 (14:23 +0000)]
- remove unused bootdelay feauture
- only print a line for every boot process if 'verbose' variable set to
nonzero; reason: with serial output, the long output
significantly slows down frequent reboots, and causes 'scroll damage'
that in some cases is pretty bad. also the verbose output doesn't tell
you the one thing you might want to know about a process: how much memory
is it using? or how much memory is everything using?
- short format does print out total memory allocated for processes
Ben Gras [Mon, 21 Sep 2009 14:23:10 +0000 (14:23 +0000)]
- VM_KERN_NOPAGEZERO feature is gone
- sys_getbiosbuffer feature is gone (from kernel; available from vm)
- bump version number because munmap() calls that newly compiled binaries
will do trigger an ugly (but harmless) error message in older VM's
- some new VM calls and flags, the new IPC calls
- some new CR0 register bits
- added files for shared memory
Tomas Hruby [Tue, 15 Sep 2009 10:01:06 +0000 (10:01 +0000)]
Some clean up of the segment selectors macros
- [ABCD]_INDEX are not used anywhere
- value of *_SELECTOR is now calculated using the *_INDEX value so changing the
index does not break the selector
- TSS is now the last of the global selectors. There will be TSS per CPU on SMP
and the number will vary depending on the maximal supported number of CPUs
configured
Tomas Hruby [Tue, 15 Sep 2009 09:57:22 +0000 (09:57 +0000)]
proc_addr() returns address based on location in proc array
- pproc_addr is not neccessary to get the address of a process if we know its
number
- local proc variables in system calls implementation (sys_task) conflicts with
the global proc array of all process, therefore the variable were renamed to
proc_nr as they hold the process number
Disabled check in test 28 which hard links a directory; this is nott required by POSIX and not supported (currently) by MINIX. Also corrected total number of tests.
Allow setuid tests 11 and 33 to run. The former still fails (but now with a meaningful error) while the latter succeeds. Only 2 tests are left borken on default MINIX, namely 11 and 28.
Thomas Veerman [Thu, 3 Sep 2009 09:38:27 +0000 (09:38 +0000)]
- Added netconf script which makes it a lot easier to change network settings.
- Modified the setup script to use the netconf script for the network
configuration:
- Moved step 2 to step 8 and renamed the steps in between.
- Autopart adapted to print step 3 instead of step 4.
Tomas Hruby [Sun, 30 Aug 2009 14:55:30 +0000 (14:55 +0000)]
system_init() renamed to arch_init()
- a better name for architecture specific init function
- some of x86 init code must execute in protected mode
- prot_init() removed from this function and still called in cstart() Imho this
should be called from the architecture specific assembly not cstart. cstart
perform Minix monitor specific tasks and will be touched once another
bootloader is in use, e.g. booting via tftp, therefore we keep it as is for
now.
- this is a backport from the SMP code which requires this. Merging will be simpler
Tomas Hruby [Sat, 29 Aug 2009 19:26:46 +0000 (19:26 +0000)]
saved_proc in exception() may be NULL
If an exception happens in kernel while the kernel is booting and no processes
are running yet, saved_proc == NULL and priting any process related information
results in dumping rubish.
This check is mostly useful when debugging kernel stuff. Should _never_ happen
on a production kernel.
Thomas Veerman [Tue, 18 Aug 2009 11:36:01 +0000 (11:36 +0000)]
Added a hack to start binaries from the boot image only. In particular, setting
bin_img=1 in the boot monitor will make sure that during the boot procedure the
mfs binary that is part of the boot image is the only binary that is used to
mount partitions. This is useful when for some reason the mfs binary on disk
malfunctions, rendering Minix unable to boot. By setting bin_img=1, the binary
on disk is ignored and the binary in the boot image is used instead.
- 'service' now accepts an additional flag -r. -r implies -c. -r instructs RS
to first look in memory if the binary has already been copied to memory and
execute that version, instead of loading the binary from disk. For example,
the first time a MFS is being started it is copied (-c) to memory and
executed from there. The second time MFS is being started this way, RS will
look in memory for a previously copied MFS binary and reuse it if it exists.
- The mount and newroot commands now accept an additional flag -i, which
instructs them to set the MS_REUSE flag in the mount flags.
- The mount system call now supports the MS_REUSE flag and invokes 'service'
with the -r flag when MS_REUSE is set.
- /etc/rc and the rc script that's included in the boot image check for the
existence of the bin_img flag in the boot monitor, and invoke mount and
newroot with the -i flag accordingly.
Thomas Veerman [Wed, 12 Aug 2009 19:57:37 +0000 (19:57 +0000)]
- Moved (u)mount prototypes from unistd.h to sys/mount.h.
- Prepared mount system call to accept multiple mount flags
instead of just read_only (however, it remains backwards
compatible).
- Updated the man mount(2) to reflect new header file usage.
- Updated badblocks, newroot, mount, and umount commands to use the
new header file.
Thomas Veerman [Tue, 14 Jul 2009 09:39:05 +0000 (09:39 +0000)]
Fixed a minor select bug:
- When one does a select on a file descriptor that is meaningless for that particular file type, select shall indicate that the file descriptor is ready for that particular operation and that the file descriptor has no exceptional condition pending.