* dump_allocations_per_caller(): Compute the total allocation count and
size from the caller infos instead of using return arguments in the
helper functions called.
* Move caller info update code from analyze_allocation_callers() to new
function slab_debug_add_allocation_for_caller(), so it can be reused.
* Add MemoryManager::AnalyzeAllocationCallers() to collect the
allocation information for the memory manager.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@43078 a95241bf-73f2-0310-859d-f6bbb57e9c96
Added a comment about the Milan clone which uses the mmu in the BIOS to emulate Atari hardware, and the Transparent Translation registers to map the PCI bus, which screws up with our current code.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@43073 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Add optional stack trace capturing for slab memory manager tracing.
* Add allocation tracking for the slab allocator (enabled via
SLAB_ALLOCATION_TRACKING). The allocation tracking requires tracing
with stack traces to be enabled for object caches and/or the memory
manager.
- Add class AllocationTrackingInfo that associates an allocation with
its respective tracing entry. The structure is added to the end of
an allocation done by the memory manager. For the object caches
there's a separate array for each slab.
- Add code range markers to the slab code, so that the first caller
into the slab code can be retrieved from the stack traces.
- Add KDL command "allocations_per_caller" that lists all allocations
summarized by caller.
* Move debug definitions from slab_private.h to slab_debug.h.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@43072 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Move struct tracing_stack_trace to tracing.h header.
* Add tracing_find_caller_in_stack_trace(). Helper function to get the
first return address of a stack trace that is not in one of the given
address ranges.
* Add AbstractTracingEntryWithStackTrace::StackTrace() getter.
* Add tracing_is_entry_valid(). Checks, based on the additionally given
time, whether a tracing entry is (probably) still in the tracing
buffer.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@43070 a95241bf-73f2-0310-859d-f6bbb57e9c96
Move blocking the 0xcccccccc and 0xdeadbeef address ranges from heap to VM init
so that it also works when used in the slab allocator.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@43047 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Introduce "paranoid" malloc/free into the slab allocator (initializing
allocated memory to 0xcc and setting freed memory to 0xdeadbeef).
* Allow for optional stack traces for slab object cache tracing.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@43046 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Add an AbstractTraceEntryWithStackTrace that includes stack trace handling.
* Add a selector macro/template combo to conveniently select the right base
class depending on whether stack traces are enabled or not.
* Minor style cleanups.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@43045 a95241bf-73f2-0310-859d-f6bbb57e9c96
instead of a mix of B_NOT_SUPPORTED and B_UNSUPPORTED. This allows checking for
a specific error code. Probably one of those should be phased out...
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@43025 a95241bf-73f2-0310-859d-f6bbb57e9c96
often enough and simple enough to write that we should allow it.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42865 a95241bf-73f2-0310-859d-f6bbb57e9c96
be used to mark certain io interrupt vectors as reserved and to allocate from
the still free ones. It is a kernel private API for now though.
* Make the MSI code use that functionality instead of implementing its own which
slims it down considerably and also removes quite a bit of hardcoded knowledge
about the interrupt layout that didn't really belong there.
* Mark the various in-use interrupts as reserved from the components that
actually know about them (PIC, IO-APIC, SMP, APIC timer and interrupt setup).
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42832 a95241bf-73f2-0310-859d-f6bbb57e9c96
has no routing information but wasn't configured by the BIOS either. The
function will now only panic if a device that was previously configured would
not be so anymore after enabling the IO-APIC. Fixes#7971.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42795 a95241bf-73f2-0310-859d-f6bbb57e9c96
to a panic at boot.
* Make the panic message more explicit when there is no more room left.
This should hopefully fix#7869.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42715 a95241bf-73f2-0310-859d-f6bbb57e9c96
for message formatting
* adjust LocaleKit to use namespace 'icu', as ICU has been configured to no longer
use a version specific namespace
* adjust LocaleKit to general API changes in ICU 4.8
Note: all software using ICU (like WebPositive) needs to be rebuilt!
Note: the ICU package for PPC needs to be updated before it can be used!
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42638 a95241bf-73f2-0310-859d-f6bbb57e9c96
directory of a file without traversing leaf links (just like lstat()).
* Minor cleanup.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42620 a95241bf-73f2-0310-859d-f6bbb57e9c96
faster than before.
* This also solves a TODO in dladdr(), although I did not use
get_library_symbol() as I didn't quite see how that could fit as the comment
suggested; there is now a new function get_symbol_at_address() for this.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42598 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Rename of_support.h/cpp back to support.cpp as per Axel
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42498 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Rename a few variables to make more sense
* OF_FAILED is a signed int.. fix return of of_address_cells
* OF_FAILED is a signed int.. fix return of of_size_cells
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42497 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Add header file to support of_support.cpp
* Add support functions to obtain address and size cell lengths
* Small style cleanups
* Add support for G5 PowerPC cpus...
* Refactor memory region code to be aware of 64-bit OF addresses.
As-is the boot loader wouldn't start on G5 systems because
OpenFirmware memory base addresses are stored as two 32-bit
unsigned int 'cells' vs one 32-bit unsigned int 'cell' on G3/G4.
I removed the static struct and replaced it with a template
and pass uint32 or uint64 depending on the address cell size.
Thanks for the idea DeadYak!
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42486 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Style fixes as per Axel
* Reintroduce removed OF_FAIL checks
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42460 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Detect OpenBIOS used in QEMU and set machine flag
(OpenBIOS isn't 1:1 Apple OpenFirmware)
* Show at boot which machine type is detected
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42459 a95241bf-73f2-0310-859d-f6bbb57e9c96
* Clean up messge and error text
* Begin use B_PRI* macros
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42449 a95241bf-73f2-0310-859d-f6bbb57e9c96
introduced in r37701 was not taken into account for computing the short meta
chunk size. Fixes the reopened#6237.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42385 a95241bf-73f2-0310-859d-f6bbb57e9c96
512 bytes with a heap page of 2048 bytes resulting in excessive waste for
allocations between 512 and 1023 bytes. Also tune the requested alignment so
that sizeof(port_message) (currently 28 bytes) can occupy one allocation unit
without waste.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42340 a95241bf-73f2-0310-859d-f6bbb57e9c96
resize operation that has to be undone and may fail when doing so.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42332 a95241bf-73f2-0310-859d-f6bbb57e9c96
we have to enlarge/shrink the array that holds them and assign a protection
value for the additional pages as necessary. Otherwise we'll access invalid
memory when looking up page protections for enlarged areas and get random
protection values.
Experienced with QEMU that sets page protections via mprotect on heap memory.
When the heap was later enlarged, write access to the additional memory would
result in permission denied errors and crashes.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42330 a95241bf-73f2-0310-859d-f6bbb57e9c96
debugger. As sInDebugger is already > 0 when the first CPU enters KDL, code
from other CPUs might see debug_debugger_running() == true already before they
enter the debugger.
* Instead, move the sDebuggerOnCPU setting out of the debugger loop and hold the
value until after calling exit_kernel_debugger() so that the exit hooks still
see debug_debugger_running() == true.
* Also avoid calling exit_kernel_debugger() when we've been called recursively
(previousCPU != -1). Previously the exit hooks would've been called and the
debugger state reset erroneously. To balance the missing decrement of
sInDebugger in that case we decrement sInDebugger in enter_kernel_debugger()
also when detecting the recursion case.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42320 a95241bf-73f2-0310-859d-f6bbb57e9c96
running. The former has a broader scope and lasts until the debugger exit is
actually done whereas the latter is already reset when the inner loop is exited.
This fixes the issue Ingo saw where the USB physical memory manager wasn't able
to free resources used for the debug transfer. It has reserved debug memory that
it uses depending on debug_debugger_running() and was therefore confused when
it returned false when called from the kernel debugger module exit hook.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42319 a95241bf-73f2-0310-859d-f6bbb57e9c96
enabled as well. As this heap implementation is still used for the port heap
(as it handles B_NO_LOCK areas) those are still useful.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42278 a95241bf-73f2-0310-859d-f6bbb57e9c96
last byte of an unmapped-but-still-there page non-readable (i.e. from B_NO_LOCK
areas), causing such reads to fail in KDL.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42276 a95241bf-73f2-0310-859d-f6bbb57e9c96
previously just always write over the beginning of the buffer for each vector.
Since the writev version isn't exposed to userland by means of a syscall and
kernel internally nobody used it, nobody noticed so far.
* Merge the two loops for user and kernel copy to remove the code duplication.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42273 a95241bf-73f2-0310-859d-f6bbb57e9c96
the acquired quota in sTotalSpaceInUse wasn't released in all cases leading to
it eventually reaching the limit (after a _very_ long time though, so this is
more theoretical than anything else). The sAllocatingArea flag wasn't reset in
the case that an area was already added in the meantime, resulting in no
further growing being possible. Then there were race conditions between
waiting for space to become available and the situations which made that space
available (freeing port_messages and adding new areas).
* Fix these race conditions by using a mutex (sPortQuotaLock) to protect the
various quota and allocation related variables. Instead removed the atomic_*
operations that were previously used.
* Had to move some static functions around.
Should make port heap growing more robust, even though in normal use you'll
likely never encounter it...
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42272 a95241bf-73f2-0310-859d-f6bbb57e9c96
APIC enumeration. As the local APIC is the basis for inter-CPU communication, we
can't do SMP when it's absent.
Fixes#7692, but indeed it makes no real sense for QEMU to provide the APIC info
in the MP and ACPI tables (and actually emulate the hardware) when it disables
the APIC support flag at the same time.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42215 a95241bf-73f2-0310-859d-f6bbb57e9c96
specified the syscall error code B_WOULD_BLOCK would no longer be mapped to 0.
Fixes#7693.
git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@42188 a95241bf-73f2-0310-859d-f6bbb57e9c96
