Basically, there was a pretty subtle race between the cpus in main where if the main cpu released the AP cpus and then before the AP cpus had a chance to run the boot cpu started creating the main thread (which causes smp ici messages to be created) the system would livelock, where the boot cpu waited forever for the AP cpu to acknowledge the ICI (for a TLB flush when creating the kernel stack).
Added smp_cpu_rendezvous(), used to synchronize all the cpus to a particular point, and used it a few times in main().
While i was at it i fixed another race that'll probably never happen, but what the hey. Make sure the kernel args are copied into kernel space by the main cpu before letting any other ones use it.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20269 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Made publish_vnode() available in userland. For old style FS add-ons
publish_vnode() is used when they request a new_vnode(). The semantics
of new_vnode() changed considerably in Haiku, but publish_vnode()
seems to do pretty much what the old new_vnode() did.
* The UserlandFS hosted RamFS begins to work under Haiku. It runs pretty
soon out of memory though (under vmware with 256 MB) and node
monitoring is broken ATM.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20264 a95241bf-73f2-0310-859d-f6bbb57e9c96
old interface is completely done in userland ATM.
It becomes more and more obvious that we probably need to provide
the kernel add-on with a bit more information about what the client FS
interface supports in the first place, so we can save unnecessary trips to
the userland. Opening/closing attributes for a FS using the old style
interface could be handled completely in the kernel add-on, for instance
(even if we lose a bit of accuracy wrt to open modes etc.).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20258 a95241bf-73f2-0310-859d-f6bbb57e9c96
need via library libuserlandfs_beos_kernel.so. Fine-tuned the legacy headers
so they can by used by the the kernel interface emulation code as well as by
the add-ons. This is actually a bit hacky, since we build everything in the
Haiku build environment and thus mix these old headers and Haiku's.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20254 a95241bf-73f2-0310-859d-f6bbb57e9c96
* vm_area and vm_page now have a new field "mappings" where they will store lists
of vm_page_mapping structures. vm_page::ref_count is gone, as it's no longer
needed (it was never updated correctly, anyway).
* vm_caches now have a type field, ie. CACHE_TYPE_RAM for anonymous areas - this
makes the stores a bit less independent, but is quite handy in several places.
* Added new vm_map_page() and vm_unmap_pages() functions to be used whenever you
map in or unmap pages into/from an area. They don't do much more than handling
vm_page::wired_count correctly right now, though (ie. B_LAZY_LOCK is now working
as expected as well).
* Moved the device fault handler to vm_map_physical_memory(); it was not really
used as a fault handler, anyway.
* Didn't notice Ingo's changes to the I/O space region broke lock_memory(). It
now checks the type of the area that contains the memory, and doesn't lock
anymore if not needed which solves the problem in a platform independent way.
* Implemented lock_memory() and unlock_memory() for real: they now change the
vm_page::wired_count member to identify pages that shouldn't be paged out.
* vm_area_for() now uses vm_area_lookup() internally.
* Fixed various potential overflow conditions with areas that reach 0xffffffff.
* Creating anonymous areas with B_FULL_LOCK no longer causes vm_soft_fault()
to be called, instead, the pages are allocated and mapped (via vm_map_page())
directly.
* Removed the _vm_ prefix for create_area_struct() and create_reserved_area_struct().
* Fixed a bug in vm_page_write_modified() that would not have enqueued pages that
failed to be written to the modified queue again when needed.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20251 a95241bf-73f2-0310-859d-f6bbb57e9c96
interface. The classes actually interfacing with the client FS add-on still
need to be adjusted.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20246 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Renamed them, made everything static besides vm_allocate_early() (previous
vm_alloc_from_kernel_args()) which now allows you to specify a different
virtual than physical size, and therefore makes vm_alloc_virtual_from_kernel_args()
superfluous (which isn't exported anymore, and is now called allocate_early_virtual()).
* Enabled printing a stack trace on serial output on team crash - it doesn't hurt
for now, anyway.
* Cleanup.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20244 a95241bf-73f2-0310-859d-f6bbb57e9c96
FS interface. Adjusted old hooks, but didn't add the new ones yet. The
module builds now at least.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20235 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Renamed a few parameters of the FS module hooks:
- *file in the attribute functions to *vnode
- v to vnode
- I could barely restrain myself from renaming the "_*" parameters. I
understand this marks return parameters, but I'd prefer a nicer prefix
or suffix (that doesn't makes you think this is a private/internal
identifier) like "out", "ret", "return", "result", or something similar.
I'd also like to propose renaming {read,write}_link() to
{read,write}_symlink(). Er, and do we need write_link() at all?
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20226 a95241bf-73f2-0310-859d-f6bbb57e9c96
This fixes bug #242. The value is currently stored in a separate file.
* Removed some unused codes from ServerProtocol.h.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20188 a95241bf-73f2-0310-859d-f6bbb57e9c96
The first use is to let the kernel decide what the preferred syscall mechanism is at boot time and copy the
appropriate user space code there. Can be used for routines the kernel can decide best how to use (memcpy, some
timing routines, etc).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20161 a95241bf-73f2-0310-859d-f6bbb57e9c96
the new cpuid stuff was apparently exacerbating an existing problem where various bits of low level
cpu code (specifically get_current_cpu) weren't really initialized before being used. Changed the
order to set up a fake set of threads to point each cpu at really early in boot to make sure that at
all points in code it can get the current 'thread' and thus the current cpu.
A probably better solution would be to have dr3 point to the current cpu which would then point to the
current thread, but that has a race condition that would require an int disable, etc.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20160 a95241bf-73f2-0310-859d-f6bbb57e9c96
updated the mesa software addon to work again (maybe would need more work)
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20137 a95241bf-73f2-0310-859d-f6bbb57e9c96
Now two complete tss structures exist within the per-cpu structure. Instead
of having to create a seperate area per each one, initialize them in place.
Also, the old mechanism to getting all of the cpus to get initialized was
subtly broken, but still managed to work. Now, just force all the cpus to
initialize at boot, which makes the actual swapping of esp0 somewhat simpler.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20131 a95241bf-73f2-0310-859d-f6bbb57e9c96
* recursive_lock_unlock() now returns a void to mirror it's counterpart better;
use recursive_lock_get_recursion() if you're interested in the lock depth.
* switch_sem(), and release_sem() now don't do anything anymore in kernel startup
mode.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20099 a95241bf-73f2-0310-859d-f6bbb57e9c96
for one quantum wasn't really a good idea, as this could get quite expensive for the thread
(depending on the system load, it might have taken a long time until the thread was scheduled
again, no matter what priority it was).
Also, calling thread_yield() in a loop would have taken 100% CPU time.
Now, we sort the thread into the queue as with any other thread, but we'll ignore it once.
This now guarantees an actual context switch, as well as a much fairer rescheduling policy
for threads calling that function.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20077 a95241bf-73f2-0310-859d-f6bbb57e9c96
at boot, per cpu, detect the cpu, pull down all the relevant cpuid bits and
save them into the per-cpu structure. Changed most of the code scattered here
and there that reads the cpuid to use a new api, x86_check_feature, which looks
at the saved bits.
Also changed the system_info stuff to read from these bits.
While i was at it, refreshed all the bits to be current.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20072 a95241bf-73f2-0310-859d-f6bbb57e9c96
Replaced the _kern_null syscall with _kern_is_computer_on.
is_computer_on_fire is a bit harder, since it returns a float from kernelland, which
at the moment isn't supported in haiku.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20069 a95241bf-73f2-0310-859d-f6bbb57e9c96
* replaced new[] with malloc()/realloc() where appropriate - since we're messing with
the bits anyway, this makes the code slightly faster.
* however, we might want to throw some std::bad_alloc exceptions to deal correctly with
low memory situations.
* cleanup.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20065 a95241bf-73f2-0310-859d-f6bbb57e9c96
this fixes bug #1005. As a result, the Disks icon will now appear in file panels
when you change that setting with a panel open.
* _ObserverList is now in the BPrivate namespace (and renamed to ObserverList).
* its BHandler map is now only temporarily used for handlers that do not belong to
a looper yet; when BHandler::SendNotices() is called, they will be transferred
into the BMessenger map.
* Invalid messengers are now removed from the map when encountered.
* Added TODO comments about a possible reference counting if a handler is added
twice to a list (right now, this will be ignored).
* All {Start|Stop}Watching() methods are now more or less safe to be used in low
memory situations (adding some items to the map can still throw an exception...).
* Renamed BHandler::InitData() to _InitData().
* Some refactoring and cleanup.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20029 a95241bf-73f2-0310-859d-f6bbb57e9c96
* we now always flush the TLBs after having unmapped some pages.
* vm_soft_fault() could traverse to a source cache while it was being collapsed
by vm_cache_remove_consumer() - this is now no longer possible as the latter
marks the cache as busy when doing so, and the former now tests this flag and
locks the cache (via the new fault_acquire_locked_source() function).
* if fault_acquire_locked_source() fails with B_BUSY, the current cache is locked
again, and tested again for the page - as it might have been moved upwards to it
with the destruction of its former source.
* The cache delivering the page for vm_soft_fault() is now locked until the end;
it can no longer go away before having actually mapped the page into the area.
* This also fixes the issue where pages would get lost as vm_soft_fault() put the
page in the active list, no matter if its cache still existed.
* Also, we now keep a reference of to a cache in case a dummy page is inserted; this
makes again sure that it doesn't go away during the execution of vm_soft_fault()
(which could even add this page to the free list...).
* divided vm_soft_fault() into several smaller functions which should make it much
more readable.
* Added a "cache_chain" KDL command that dumps the whole chain until the bottom
when giving a pointer to a vm_cache as parameter.
* now usually call vm_cache_acquire_ref() before map_backing_store(), even though
it shouldn't be really needed (I added it for debugging purposes).
* Some minor cleanup.
* NOTE: a major problem still persists: when removing a vm_cache, it's possible
that some of its pages are still mapped, and there is currently no mechanism
to get rid of these mappings! I've added TODO comments into vm_cache.c where
appropriate.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20028 a95241bf-73f2-0310-859d-f6bbb57e9c96
process group. This fixes bug #996.
* As a result, the process group stuff and wait_for_child() got much simpler;
get_death_entry() and update_wait_for_any() could go away completely.
* If a team goes away, all of its children are now "reparented" to the kernel team,
instead of the team's parent - this follows common implementations (and POSIX if
I understand it correctly), but not BeOS anymore. The OpenGroup Base says this
about this topic: "If a parent process terminates without waiting for all of its
child processes to terminate, the remaining child processes shall be assigned a
new parent process ID corresponding to an implementation-defined system process."
* We wait too long in wait_test_4 which at least puts us on par with Linux; see
comment in _user_setpgid().
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@20010 a95241bf-73f2-0310-859d-f6bbb57e9c96
C style ones, as this is a C header, too (and a very basic one).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@19984 a95241bf-73f2-0310-859d-f6bbb57e9c96
feel free to change that ;-)
* Cleaned up existing headers.
* Coding style guide update to BBufferIO (renamed m_* members to f*).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@19972 a95241bf-73f2-0310-859d-f6bbb57e9c96
with PostMessage() in case the message queue is full.
Some notes:
* for synchronous replies, we don't use this mechanism yet, but it could be
extended to do that as well.
* the code looks so complicated because we need a way to access the looper's
queue without locking it (to prevent deadlocks); like Dano's solution, I've
abused BTokenSpace to store a BDirectMessageTarget with a BHandler.
* we also need to decouple the lifetime of a looper's queue from its target,
as we cannot lock the looper, and therefore, can't guarantee it stays valid
as long as we're accessing it outside of BLooper.
* init_clipboard() now needs to be done after the global constructors have
been called - since sending messages now needs gDefaultTokens to be initialized.
Since this is done per image, it shouldn't cause any troubles, though.
* some minor cleanup, removed unused _msg_cache_cleanup_() and friends.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@19968 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Got rid of unused BLooper members
* renamed fTaskID to fThread
* Removed private and deprecated AddLooper()/RemoveLooper()/... stuff; BLooper is now
directly calling BLooperList methods.
* Got rid of extensive and useless comments
* Made a few TODOs more clear
* Merged InitData() and InitData(...) to _InitData(...)
* BLooper::Team() now uses BPrivate::current_team(), sTeamID is gone now.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@19966 a95241bf-73f2-0310-859d-f6bbb57e9c96
* BLooper::AddMessage() is now using this method to determine wether or not to
update its looper thread.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@19956 a95241bf-73f2-0310-859d-f6bbb57e9c96
