Commit graph

33 commits

Author SHA1 Message Date
Ben Gras
ddde360e3e vm - hash table for block cache 2010-10-15 09:10:14 +00:00
Cristiano Giuffrida
0d984b36ef Get rid of static spare pages after VM startup. 2010-07-20 21:59:27 +00:00
Ben Gras
86139e8bf2 vm: last commit should not have included wiping out these calls
(this was done temporarily to make vm work with clang; vm crashes when
the 64-bit arithmetic functions are used when compiled with clang.)
2010-07-05 14:01:34 +00:00
Ben Gras
b4bea1bfcb vm: fixes for clang warnings 2010-07-05 13:58:57 +00:00
Cristiano Giuffrida
06700d05d1 Give RS a page table. 2010-06-28 21:53:37 +00:00
Arun Thomas
b641afc78a VM: Remove legacy non-paging code paths 2010-06-05 14:39:40 +00:00
Erik van der Kouwe
1f11a57141 Oops, last commit included more than was intended 2010-05-20 08:07:47 +00:00
Erik van der Kouwe
5f15ec05b2 More system processes, this was not enough for the release script to run on some configurations 2010-05-20 08:05:07 +00:00
Ben Gras
f78d8e74fd secondary cache feature in vm.
A new call to vm lets processes yield a part of their memory to vm,
together with an id, getting newly allocated memory in return. vm is
allowed to forget about it if it runs out of memory. processes can ask
for it back using the same id. (These two operations are normally
combined in a single call.)

It can be used as a as-big-as-memory-will-allow block cache for
filesystems, which is how mfs now uses it.
2010-05-05 11:35:04 +00:00
Tomas Hruby
f51eea4b32 Changed pagefault delivery to VM
this patch changes the way pagefaults are delivered to VM. It adopts
the same model as the out-of-quantum messages sent by kernel to a
scheduler.

- everytime a userspace pagefault occurs, kernel creates a message
  which is sent to VM on behalf of the faulting process

- the process is blocked on delivery to VM in the standard IPC code
  instead of waiting in a spacial in-kernel queue (stack) and is not
  runnable until VM tell kernel that the pagefault is resolved and is
  free to clear the RTS_PAGEFAULT flag.

- VM does not need call kernel and poll the pagefault information
  which saves many (1/2?) calls and kernel calls that return "no more
  data"

- VM notification by kernel does not need to use signals

- each entry in proc table is by 12 bytes smaller (~3k save)
2010-04-26 23:21:26 +00:00
Ben Gras
c78250332d let vm use physically fragmented memory for allocations.
map_copy_ph_block is replaced by map_clone_ph_block, which can
replace a single physical block by multiple physical blocks.

also,
 . merge map_mem.c with region.c, as they manipulate the same
   data structures
 . NOTRUNNABLE removed as sanity check
 . use direct functions for ALLOC_MEM and FREE_MEM again
 . add some checks to shared memory mapping code
 . fix for data structure integrity when using shared memory
 . fix sanity checks
2010-04-12 11:25:24 +00:00
Arun Thomas
4ed3a0cf3a Convert kernel over to bsdmake 2010-04-01 22:22:33 +00:00
Ben Gras
ec30f25d0c VM: fix kernel mappings for children of non-paged parents. 2010-03-18 17:17:31 +00:00
Cristiano Giuffrida
cb176df60f New RS and new signal handling for system processes.
UPDATING INFO:
20100317:
        /usr/src/etc/system.conf updated to ignore default kernel calls: copy
        it (or merge it) to /etc/system.conf.
        The hello driver (/dev/hello) added to the distribution:
        # cd /usr/src/commands/scripts && make clean install
        # cd /dev && MAKEDEV hello

KERNEL CHANGES:
- Generic signal handling support. The kernel no longer assumes PM as a signal
manager for every process. The signal manager of a given process can now be
specified in its privilege slot. When a signal has to be delivered, the kernel
performs the lookup and forwards the signal to the appropriate signal manager.
PM is the default signal manager for user processes, RS is the default signal
manager for system processes. To enable ptrace()ing for system processes, it
is sufficient to change the default signal manager to PM. This will temporarily
disable crash recovery, though.
- sys_exit() is now split into sys_exit() (i.e. exit() for system processes,
which generates a self-termination signal), and sys_clear() (i.e. used by PM
to ask the kernel to clear a process slot when a process exits).
- Added a new kernel call (i.e. sys_update()) to swap two process slots and
implement live update.

PM CHANGES:
- Posix signal handling is no longer allowed for system processes. System
signals are split into two fixed categories: termination and non-termination
signals. When a non-termination signaled is processed, PM transforms the signal
into an IPC message and delivers the message to the system process. When a
termination signal is processed, PM terminates the process.
- PM no longer assumes itself as the signal manager for system processes. It now
makes sure that every system signal goes through the kernel before being
actually processes. The kernel will then dispatch the signal to the appropriate
signal manager which may or may not be PM.

SYSLIB CHANGES:
- Simplified SEF init and LU callbacks.
- Added additional predefined SEF callbacks to debug crash recovery and
live update.
- Fixed a temporary ack in the SEF init protocol. SEF init reply is now
completely synchronous.
- Added SEF signal event type to provide a uniform interface for system
processes to deal with signals. A sef_cb_signal_handler() callback is
available for system processes to handle every received signal. A
sef_cb_signal_manager() callback is used by signal managers to process
system signals on behalf of the kernel.
- Fixed a few bugs with memory mapping and DS.

VM CHANGES:
- Page faults and memory requests coming from the kernel are now implemented
using signals.
- Added a new VM call to swap two process slots and implement live update.
- The call is used by RS at update time and in turn invokes the kernel call
sys_update().

RS CHANGES:
- RS has been reworked with a better functional decomposition.
- Better kernel call masks. com.h now defines the set of very basic kernel calls
every system service is allowed to use. This makes system.conf simpler and
easier to maintain. In addition, this guarantees a higher level of isolation
for system libraries that use one or more kernel calls internally (e.g. printf).
- RS is the default signal manager for system processes. By default, RS
intercepts every signal delivered to every system process. This makes crash
recovery possible before bringing PM and friends in the loop.
- RS now supports fast rollback when something goes wrong while initializing
the new version during a live update.
- Live update is now implemented by keeping the two versions side-by-side and
swapping the process slots when the old version is ready to update.
- Crash recovery is now implemented by keeping the two versions side-by-side
and cleaning up the old version only when the recovery process is complete.

DS CHANGES:
- Fixed a bug when the process doing ds_publish() or ds_delete() is not known
by DS.
- Fixed the completely broken support for strings. String publishing is now
implemented in the system library and simply wraps publishing of memory ranges.
Ideally, we should adopt a similar approach for other data types as well.
- Test suite fixed.

DRIVER CHANGES:
- The hello driver has been added to the Minix distribution to demonstrate basic
live update and crash recovery functionalities.
- Other drivers have been adapted to conform the new SEF interface.
2010-03-17 01:15:29 +00:00
Arun Thomas
1f9ce647cf Move archtypes.h, fpu.h, and stackframe.h
Move archtypes.h to include/ dir, since several servers require it. Move
fpu.h and stackframe.h to arch-specific header directory. Make source
files and makefiles aware of the new header locations.
2010-03-09 09:41:14 +00:00
Ben Gras
35a108b911 panic() cleanup.
this change
   - makes panic() variadic, doing full printf() formatting -
     no more NO_NUM, and no more separate printf() statements
     needed to print extra info (or something in hex) before panicing
   - unifies panic() - same panic() name and usage for everyone -
     vm, kernel and rest have different names/syntax currently
     in order to implement their own luxuries, but no longer
   - throws out the 1st argument, to make source less noisy.
     the panic() in syslib retrieves the server name from the kernel
     so it should be clear enough who is panicing; e.g.
         panic("sigaction failed: %d", errno);
     looks like:
         at_wini(73130): panic: sigaction failed: 0
         syslib:panic.c: stacktrace: 0x74dc 0x2025 0x100a
   - throws out report() - printf() is more convenient and powerful
   - harmonizes/fixes the use of panic() - there were a few places
     that used printf-style formatting (didn't work) and newlines
     (messes up the formatting) in panic()
   - throws out a few per-server panic() functions
   - cleans up a tie-in of tty with panic()

merging printf() and panic() statements to be done incrementally.
2010-03-05 15:05:11 +00:00
David van Moolenbroek
61bb82a44b VM information interface 2010-01-19 21:00:20 +00:00
Cristiano Giuffrida
c5b309ff07 Merge of Wu's GSOC 09 branch (src.20090525.r4372.wu)
Main changes:
- COW optimization for safecopy.
- safemap, a grant-based interface for sharing memory regions between processes.
- Integration with safemap and complete rework of DS, supporting new data types
  natively (labels, memory ranges, memory mapped ranges).
- For further information:
  http://wiki.minix3.org/en/SummerOfCode2009/MemoryGrants

Additional changes not included in the original Wu's branch:
- Fixed unhandled case in VM when using COW optimization for safecopy in case
  of a block that has already been shared as SMAP.
- Better interface and naming scheme for sys_saferevmap and ds_retrieve_map
  calls.
- Better input checking in syslib: check for page alignment when creating
  memory mapping grants.
- DS notifies subscribers when an entry is deleted.
- Documented the behavior of indirect grants in case of memory mapping.
- Test suite in /usr/src/test/safeperf|safecopy|safemap|ds/* reworked
  and extended.
- Minor fixes and general cleanup.
- TO-DO: Grant ids should be generated and managed the way endpoints are to make
sure grant slots are never misreused.
2010-01-14 15:24:16 +00:00
Cristiano Giuffrida
d1fd04e72a Initialization protocol for system services.
SYSLIB CHANGES:
- SEF framework now supports a new SEF Init request type from RS. 3 different
callbacks are available (init_fresh, init_lu, init_restart) to specify
initialization code when a service starts fresh, starts after a live update,
or restarts.

SYSTEM SERVICE CHANGES:
- Initialization code for system services is now enclosed in a callback SEF will
automatically call at init time. The return code of the callback will
tell RS whether the initialization completed successfully.
- Each init callback can access information passed by RS to initialize. As of
now, each system service has access to the public entries of RS's system process
table to gather all the information required to initialize. This design
eliminates many existing or potential races at boot time and provides a uniform
initialization interface to system services. The same interface will be reused
for the upcoming publish/subscribe model to handle dynamic 
registration / deregistration of system services.

VM CHANGES:
- Uniform privilege management for all system services. Every service uses the
same call mask format. For boot services, VM copies the call mask from init
data. For dynamic services, VM still receives the call mask via rs_set_priv
call that will be soon replaced by the upcoming publish/subscribe model.

RS CHANGES:
- The system process table has been reorganized and split into private entries
and public entries. Only the latter ones are exposed to system services.
- VM call masks are now entirely configured in rs/table.c
- RS has now its own slot in the system process table. Only kernel tasks and
user processes not included in the boot image are now left out from the system
process table.
- RS implements the initialization protocol for system services.
- For services in the boot image, RS blocks till initialization is complete and
panics when failure is reported back. Services are initialized in their order of
appearance in the boot image priv table and RS blocks to implements synchronous
initialization for every system service having the flag SF_SYNCH_BOOT set.
- For services started dynamically, the initialization protocol is implemented
as though it were the first ping for the service. In this case, if the
system service fails to report back (or reports failure), RS brings the service
down rather than trying to restart it.
2010-01-08 01:20:42 +00:00
David van Moolenbroek
ac9ab099c8 General cleanup:
- clean up kernel section of minix/com.h somewhat
- remove ALLOCMEM and VM_ALLOCMEM calls
- remove non-safecopy and minix-vmd support from Inet
- remove SYS_VIRVCOPY and SYS_PHYSVCOPY calls
- remove obsolete segment encoding in SYS_SAFECOPY*
- remove DEVCTL call, svrctl(FSDEVUNMAP), map_driverX
- remove declarations of unimplemented svrctl requests
- remove everything related to swapping to disk
- remove floppysetup.sh
- remove traces of rescue device
- update DESCRIBE.sh with new devices
- some other small changes
2010-01-05 19:39:27 +00:00
Cristiano Giuffrida
1f5841c8ed Basic System Event Framework (SEF) with ping and live update.
SYSLIB CHANGES:
- SEF must be used by every system process and is thereby part of the system
library.
- The framework provides a receive() interface (sef_receive) for system
processes to automatically catch known system even messages and process them.
- SEF provides a default behavior for each type of system event, but allows
system processes to register callbacks to override the default behavior.
- Custom (local to the process) or predefined (provided by SEF) callback
implementations can be registered to SEF.
- SEF currently includes support for 2 types of system events:
  1. SEF Ping. The event occurs every time RS sends a ping to figure out
  whether a system process is still alive. The default callback implementation
  provided by SEF is to notify RS back to let it know the process is alive
  and kicking.
  2. SEF Live update. The event occurs every time RS sends a prepare to update
  message to let a system process know an update is available and to prepare
  for it. The live update support is very basic for now. SEF only deals with
  verifying if the prepare state can be supported by the process, dumping the
  state for debugging purposes, and providing an event-driven programming
  model to the process to react to state changes check-in when ready to update.
- SEF should be extended in the future to integrate support for more types of
system events. Ideally, all the cross-cutting concerns should be integrated into
SEF to avoid duplicating code and ease extensibility. Examples include:
  * PM notify messages primarily used at shutdown.
  * SYSTEM notify messages primarily used for signals.
  * CLOCK notify messages used for system alarms.
  * Debug messages. IS could still be in charge of fkey handling but would
  forward the debug message to the target process (e.g. PM, if the user
  requested debug information about PM). SEF would then catch the message and
  do nothing unless the process has registered an appropriate callback to
  deal with the event. This simplifies the programming model to print debug
  information, avoids duplicating code, and reduces the effort to print
  debug information.

SYSTEM PROCESSES CHANGES:
- Every system process registers SEF callbacks it needs to override the default
system behavior and calls sef_startup() right after being started.
- sef_startup() does almost nothing now, but will be extended in the future to
support callbacks of its own to let RS control and synchronize with every
system process at initialization time.
- Every system process calls sef_receive() now rather than receive() directly,
to let SEF handle predefined system events.

RS CHANGES:
- RS supports a basic single-component live update protocol now, as follows:
  * When an update command is issued (via "service update *"), RS notifies the
  target system process to prepare for a specific update state.
  * If the process doesn't respond back in time, the update is aborted.
  * When the process responds back, RS kills it and marks it for refreshing.
  * The process is then automatically restarted as for a buggy process and can
  start running again.
  * Live update is currently prototyped as a controlled failure.
2009-12-21 14:12:21 +00:00
Ben Gras
f0db9bb328 - map in as much memory as is necessary in 4MB chunks to
let boot processes run with segments
 - allow segment-only processes to fork() by copying them
   and giving them an identity page table
2009-12-07 12:10:44 +00:00
Ben Gras
7e73260cf5 - enable remembering of device memory ranges set by PCI and
told to kernel
  - makes VM ask the kernel if a certain process is allowed
    to map in a range of physical memory (VM rounds it to page
    boundaries afterwards - but it's impossible to map anything
    smaller otherwise so I assume this is safe, i.e. there won't
    be anything else in that page; certainly no regular memory)
  - VM permission check cleanup (no more hardcoded calls, less
    hardcoded logic, more readable main loop), a loose end left
    by GQ
  - remove do_copy warning, as the ipc server triggers this but
    it's no more harmful than the special cases already excluded
    explicitly (VFS, PM, etc).
2009-11-03 11:12:23 +00:00
Ben Gras
32fbbd370c - pages that points to page directory values of all processes,
shared with the kernel, mapped into kernel address space; 
   kernel is notified of its location. kernel segment size is
   increased to make it fit.
 - map in kernel and other processes that don't have their
   own page table using single 4MB (global) mapping.
 - new sanity check facility: objects that are allocated with
   the slab allocator are, when running with sanity checking on,
   marked readonly until they are explicitly unlocked using the USE()
   macro.
 - another sanity check facility: collect all uses of memory and
   see if they don't overlap with (a) eachother and (b) free memory
 - own munmap() and munmap_text() functions.
 - exec() recovers from out-of-memory conditions properly now; this
   solves some weird exec() behaviour
 - chew off memory from the same side of the chunk as where we
   start scanning, solving some memory fragmentation issues
 - use avl trees for freelist and phys_ranges in regions
 - implement most useful part of munmap()
 - remap() stuff is GQ's for shared memory
2009-09-21 14:49:49 +00:00
Ben Gras
bdab3c4cfb Library call for cpu features; make kernel and vm use this to query cpu
features (specifically: 4MB pages and TLB global bit).  Only enable
these features in CR4 if available. 4MB pages to be used in the near
future.
2009-05-15 17:07:36 +00:00
Ben Gras
581e68433a basic sparepage optimisation
some simplification of linear/virtual address translation
(less hardcoding and more use of arch_*2* functions)
2009-05-12 11:38:29 +00:00
Ben Gras
d2caeb6146 start all processes, including VM, in VM_PROCSTART in linear address space,
to make space for kernel to be able to map in things below there.
2009-05-11 19:11:37 +00:00
Ben Gras
2dd02cc560 mark pages whose refcount were >1 and drop to 1 and are
read/write writable in the pagetable right away instead of waiting for
a pagefault. minor optimization.

some a sanity check of SLAB-allocated pointers.

vm gets its own _exit and __exit like PM, so the stock (library) panic works.
2009-04-22 12:39:29 +00:00
Ben Gras
3121eec6bd . map text (kernel's and processes') in readonly
. map kernel in non-user
 . don't map in first pages of kernel code and data
   if possible

these first pages could actually be freed but as the
kernel isn't allowed to touch them either we can't reuse
them until VM has totally taken over page table management
and kernel doesn't rely on identity mapping any more.
2008-12-18 15:35:22 +00:00
Ben Gras
e4e3995fb0 don't force vm to print to serial; don't kill processes when they
have 'bad' memory ranges (as it's the requestor's fault)
2008-12-11 17:36:33 +00:00
Ben Gras
68d0c4defe - code shared with exec() letting boot-time processes have
their own fully fledged virtual address space and freeing
   their pre-allocated heap+stack area (necessary to let memory
   driver map in arbitrary areas of memory for /dev/mem without
   sys_vm_map)
 - small optimization preallocating memory on exec
 - finished VR_DIRECT physical mapping code
2008-12-08 16:43:20 +00:00
Ben Gras
6c92081a5a paged mode is default. 2008-11-19 13:19:37 +00:00
Ben Gras
c078ec0331 Basic VM and other minor improvements.
Not complete, probably not fully debugged or optimized.
2008-11-19 12:26:10 +00:00