genode/repos/base/src/core/region_map_component.cc

721 lines
21 KiB
C++

/*
* \brief Implementation of the region map
* \author Christian Helmuth
* \author Norman Feske
* \author Alexander Boettcher
* \date 2006-07-17
*/
/*
* Copyright (C) 2006-2019 Genode Labs GmbH
*
* This file is part of the Genode OS framework, which is distributed
* under the terms of the GNU Affero General Public License version 3.
*/
/* Genode includes */
#include <base/log.h>
#include <base/lock.h>
#include <util/arg_string.h>
#include <util/misc_math.h>
/* core includes */
#include <util.h>
#include <cpu_session_component.h>
#include <region_map_component.h>
#include <dataspace_component.h>
static const bool verbose = false;
static const bool verbose_page_faults = false;
struct Genode::Region_map_component::Fault_area
{
addr_t _fault_addr = 0;
addr_t _base = 0;
size_t _size_log2 = 0;
addr_t _upper_bound() const {
return (_size_log2 == ~0UL) ? ~0UL : (_base + (1UL << _size_log2) - 1); }
/**
* Default constructor, constructs invalid fault area
*/
Fault_area() { }
/**
* Constructor, fault area spans the maximum address-space size
*/
Fault_area(addr_t fault_addr) :
_fault_addr(fault_addr), _size_log2(~0UL) { }
/**
* Constrain fault area to specified region
*/
void constrain(addr_t region_base, size_t region_size)
{
/*
* Find flexpage around _fault_addr that lies within the
* specified region.
*
* Start with a 'size_log2' of one less than the minimal
* page size. If the specified constraint conflicts with
* the existing fault area, the loop breaks at the first
* iteration and we can check for this condition after the
* loop.
*/
size_t size_log2 = get_page_size_log2() - 1;
addr_t base = 0;
for (size_t try_size_log2 = get_page_size_log2();
try_size_log2 < sizeof(addr_t)*8 ; try_size_log2++) {
addr_t fpage_mask = ~((1UL << try_size_log2) - 1);
addr_t try_base = _fault_addr & fpage_mask;
/* check lower bound of existing fault area */
if (try_base < _base)
break;
/* check against upper bound of existing fault area */
if (try_base + (1UL << try_size_log2) - 1 > _upper_bound())
break;
/* check against lower bound of region */
if (try_base < region_base)
break;
/* check against upper bound of region */
if (try_base + (1UL << try_size_log2) - 1 > region_base + region_size - 1)
break;
/* flexpage is compatible with fault area, use it */
size_log2 = try_size_log2;
base = try_base;
}
/* if constraint is compatible with the fault area, invalidate */
if (size_log2 < get_page_size_log2()) {
_size_log2 = 0;
_base = 0;
} else {
_size_log2 = size_log2;
_base = base;
}
}
/**
* Constrain fault area to specified flexpage size
*/
void constrain(size_t size_log2)
{
if (size_log2 >= _size_log2)
return;
_base = _fault_addr & ~((1UL << size_log2) - 1);
_size_log2 = size_log2;
}
/**
* Determine common flexpage size compatible with specified fault areas
*/
static size_t common_size_log2(Fault_area const &a1, Fault_area const &a2)
{
/*
* We have to make sure that the offset of page-fault address
* relative to the flexpage base is the same for both fault areas.
* This condition is met by the flexpage size equal to the number
* of common least-significant bits of both offsets.
*/
size_t const diff = (a1.fault_addr() - a1.base())
^ (a2.fault_addr() - a2.base());
/*
* Find highest clear bit in 'diff', starting from the least
* significant candidate. We can skip all bits lower then
* 'get_page_size_log2()' because they are not relevant as
* flexpage size (and are always zero).
*/
size_t n = get_page_size_log2();
size_t const min_size_log2 = min(a1._size_log2, a2._size_log2);
for (; n < min_size_log2 && !(diff & (1UL << n)); n++);
return n;
}
addr_t fault_addr() const { return _fault_addr; }
addr_t base() const { return _base; }
bool valid() const { return _size_log2 > 0; }
};
using namespace Genode;
static void print_page_fault(char const *msg,
addr_t pf_addr,
addr_t pf_ip,
Region_map::State::Fault_type pf_type,
Pager_object const &obj)
{
log(msg, " (",
pf_type == Region_map::State::WRITE_FAULT ? "WRITE" :
pf_type == Region_map::State::READ_FAULT ? "READ" : "EXEC",
" pf_addr=", Hex(pf_addr), " pf_ip=", Hex(pf_ip), " from ", obj, ") ");
}
/***********************
** Region-map client **
***********************/
/*
* This code is executed by the page-fault handler thread.
*/
int Rm_client::pager(Ipc_pager &pager)
{
Region_map::State::Fault_type pf_type = pager.write_fault() ? Region_map::State::WRITE_FAULT
: Region_map::State::READ_FAULT;
if (pager.exec_fault())
pf_type = Region_map::State::EXEC_FAULT;
addr_t pf_addr = pager.fault_addr();
addr_t pf_ip = pager.fault_ip();
if (verbose_page_faults)
print_page_fault("page fault", pf_addr, pf_ip, pf_type, *this);
auto lambda = [&] (Region_map_component *region_map,
Rm_region *region,
addr_t const ds_offset,
addr_t const region_offset,
addr_t const dst_region_size) -> int
{
Dataspace_component * dsc = region ? &region->dataspace() : nullptr;
if (!dsc) {
/*
* We found no attachment at the page-fault address and therefore have
* to reflect the page fault as region-manager fault. The signal
* handler is then expected to request the state of the region map.
*/
/* print a warning if it's no managed-dataspace */
if (region_map == &member_rm())
print_page_fault("no RM attachment", pf_addr, pf_ip,
pf_type, *this);
/* register fault at responsible region map */
if (region_map)
region_map->fault(*this, pf_addr - region_offset, pf_type);
/* there is no attachment return an error condition */
return 1;
}
/*
* Check if dataspace is compatible with page-fault type
*/
if (pf_type == Region_map::State::WRITE_FAULT &&
(!region->write() || !dsc->writable())) {
print_page_fault("attempted write at read-only memory",
pf_addr, pf_ip, pf_type, *this);
/* register fault at responsible region map */
if (region_map)
region_map->fault(*this, pf_addr - region_offset, pf_type);
return 2;
}
if (pf_type == Region_map::State::EXEC_FAULT) {
print_page_fault("attempted exec at non-executable memory",
pf_addr, pf_ip, pf_type, *this);
/* register fault at responsible region map */
if (region_map)
region_map->fault(*this, pf_addr - region_offset, pf_type);
return 3;
}
Mapping mapping = Region_map_component::create_map_item(region_map,
*region,
ds_offset,
region_offset,
*dsc, pf_addr,
dst_region_size);
/*
* On kernels with a mapping database, the 'dsc' dataspace is a leaf
* dataspace that corresponds to a virtual address range within core. To
* prepare the answer for the page fault, we make sure that this range is
* locally mapped in core. On platforms that support map operations of
* pages that are not locally mapped, the 'map_core_local' function may be
* empty.
*/
if (!dsc->io_mem())
mapping.prepare_map_operation();
/* answer page fault with a flex-page mapping */
pager.set_reply_mapping(mapping);
return 0;
};
return member_rm().apply_to_dataspace(pf_addr, lambda);
}
/*************
** Faulter **
*************/
void Rm_faulter::fault(Region_map_component &faulting_region_map,
Region_map::State fault_state)
{
Lock::Guard lock_guard(_lock);
_faulting_region_map = faulting_region_map.weak_ptr();
_fault_state = fault_state;
_pager_object.unresolved_page_fault_occurred();
}
void Rm_faulter::dissolve_from_faulting_region_map(Region_map_component &caller)
{
/* serialize access */
Lock::Guard lock_guard(_lock);
enum { DO_LOCK = true };
if (caller.equals(_faulting_region_map)) {
caller.discard_faulter(*this, !DO_LOCK);
} else {
Locked_ptr<Region_map_component> locked_ptr(_faulting_region_map);
if (locked_ptr.valid())
locked_ptr->discard_faulter(*this, DO_LOCK);
}
_faulting_region_map = Genode::Weak_ptr<Genode::Region_map_component>();
}
void Rm_faulter::continue_after_resolved_fault()
{
Lock::Guard lock_guard(_lock);
_pager_object.wake_up();
_faulting_region_map = Genode::Weak_ptr<Genode::Region_map_component>();
_fault_state = Region_map::State();
}
/**************************
** Region-map component **
**************************/
Mapping Region_map_component::create_map_item(Region_map_component *,
Rm_region &region,
addr_t const ds_offset,
addr_t const region_offset,
Dataspace_component &dsc,
addr_t const page_addr,
addr_t const dst_region_size)
{
addr_t const ds_base = dsc.map_src_addr();
Fault_area src_fault_area(ds_base + ds_offset);
Fault_area dst_fault_area(page_addr);
src_fault_area.constrain(ds_base, dsc.size());
dst_fault_area.constrain(region_offset + region.base(), dst_region_size);
/*
* Determine mapping size compatible with source and destination,
* and apply platform-specific constraint of mapping sizes.
*/
size_t map_size_log2 = dst_fault_area.common_size_log2(dst_fault_area,
src_fault_area);
map_size_log2 = constrain_map_size_log2(map_size_log2);
src_fault_area.constrain(map_size_log2);
dst_fault_area.constrain(map_size_log2);
if (!src_fault_area.valid() || !dst_fault_area.valid())
error("invalid mapping");
return Mapping(dst_fault_area.base(), src_fault_area.base(),
dsc.cacheability(), dsc.io_mem(),
map_size_log2, region.write() && dsc.writable(),
region.executable());
};
Region_map::Local_addr
Region_map_component::attach(Dataspace_capability ds_cap, size_t size,
off_t offset, bool use_local_addr,
Region_map::Local_addr local_addr,
bool executable, bool writeable)
{
/* serialize access */
Lock::Guard lock_guard(_lock);
/* offset must be positive and page-aligned */
if (offset < 0 || align_addr(offset, get_page_size_log2()) != offset)
throw Region_conflict();
auto lambda = [&] (Dataspace_component *dsc) {
/* check dataspace validity */
if (!dsc) throw Invalid_dataspace();
size_t const off = offset;
if (off >= dsc->size())
throw Region_conflict();
if (!size)
size = dsc->size() - offset;
/* work with page granularity */
size = align_addr(size, get_page_size_log2());
/* deny creation of regions larger then the actual dataspace */
if (dsc->size() < size + offset)
throw Region_conflict();
/* allocate region for attachment */
void *attach_at = 0;
if (use_local_addr) {
switch (_map.alloc_addr(size, local_addr).value) {
case Range_allocator::Alloc_return::OUT_OF_METADATA:
throw Out_of_ram();
case Range_allocator::Alloc_return::RANGE_CONFLICT:
throw Region_conflict();
case Range_allocator::Alloc_return::OK:
attach_at = local_addr;
break;
}
} else {
/*
* Find optimal alignment for new region. First try natural alignment.
* If that is not possible, try again with successively less alignment
* constraints.
*/
size_t align_log2 = log2(size);
if (align_log2 >= sizeof(void *)*8)
align_log2 = get_page_size_log2();
for (; align_log2 >= get_page_size_log2(); align_log2--) {
/*
* Don't use an alignment higher than the alignment of the backing
* store. The backing store would constrain the mapping size
* anyway such that a higher alignment of the region is of no use.
*/
if (((dsc->map_src_addr() + offset) & ((1UL << align_log2) - 1)) != 0)
continue;
/* try allocating the align region */
Range_allocator::Alloc_return alloc_return =
_map.alloc_aligned(size, &attach_at, align_log2);
typedef Range_allocator::Alloc_return Alloc_return;
switch (alloc_return.value) {
case Alloc_return::OK: break; /* switch */
case Alloc_return::OUT_OF_METADATA: throw Out_of_ram();
case Alloc_return::RANGE_CONFLICT: continue; /* for loop */
}
break; /* for loop */
}
if (align_log2 < get_page_size_log2())
throw Region_conflict();
}
/* store attachment info in meta data */
try {
_map.construct_metadata(attach_at,
(addr_t)attach_at, size,
dsc->writable() && writeable,
*dsc, offset, *this, executable);
}
catch (Allocator_avl_tpl<Rm_region>::Assign_metadata_failed) {
error("failed to store attachment info");
throw Invalid_dataspace();
}
Rm_region &region = *_map.metadata(attach_at);
/* inform dataspace about attachment */
dsc->attached_to(region);
/* check if attach operation resolves any faulting region-manager clients */
_faulters.for_each([&] (Rm_faulter &faulter) {
if (faulter.fault_in_addr_range((addr_t)attach_at, size)) {
_faulters.remove(faulter);
faulter.continue_after_resolved_fault();
}
});
return attach_at;
};
return _ds_ep.apply(ds_cap, lambda);
}
addr_t Region_map_component::_core_local_addr(Rm_region & region)
{
/**
* If this region references a managed dataspace,
* we have to recursively request the core-local address
*/
if (region.dataspace().sub_rm().valid()) {
auto lambda = [&] (Region_map_component * rmc) -> addr_t
{
/**
* It is possible that there is no dataspace attached
* inside the managed dataspace, in that case return zero.
*/
Rm_region * r = rmc ? rmc->_map.metadata((void*)region.offset())
: nullptr;;
return r ? rmc->_core_local_addr(*r) : 0;
};
return _session_ep.apply(region.dataspace().sub_rm(), lambda);
}
/* return core-local address of dataspace + region offset */
return region.dataspace().core_local_addr() + region.offset();
}
void Region_map_component::_unmap_region(addr_t base, size_t size)
{
if (address_space()) address_space()->flush(base, size, { 0 });
/**
* Iterate over all regions that reference this region map
* as managed dataspace
*/
for (Rm_region * r = dataspace_component().regions().first();
r; r = r->List<Rm_region>::Element::next()) {
/**
* Check whether the region referencing the managed dataspace
* and the region to unmap overlap
*/
addr_t ds_base = max((addr_t)r->offset(), base);
addr_t ds_end = min((addr_t)r->offset()+r->size(), base+size);
size_t ds_size = ds_base < ds_end ? ds_end - ds_base : 0;
/* if size is not zero, there is an overlap */
if (ds_size)
r->rm()._unmap_region(r->base() + ds_base - r->offset(), ds_size);
}
}
void Region_map_component::detach(Local_addr local_addr)
{
/* serialize access */
Lock::Guard lock_guard(_lock);
/* read meta data for address */
Rm_region *region_ptr = _map.metadata(local_addr);
if (!region_ptr) {
if (_diag.enabled)
warning("detach: no attachment at ", (void *)local_addr);
return;
}
if ((region_ptr->base() != static_cast<addr_t>(local_addr)) && _diag.enabled)
warning("detach: ", static_cast<void *>(local_addr), " is not "
"the beginning of the region ", Hex(region_ptr->base()));
Dataspace_component &dsc = region_ptr->dataspace();
/* inform dataspace about detachment */
dsc.detached_from(*region_ptr);
/*
* Create local copy of region data because the '_map.metadata' of the
* region will become unavailable as soon as we call '_map.free' below.
*/
Rm_region region = *region_ptr;
/*
* We unregister the region from region map prior unmapping the pages to
* make sure that page faults occurring immediately after the unmap
* refer to an empty region not to the dataspace, which we just removed.
*/
_map.free(reinterpret_cast<void *>(region.base()));
if (!platform().supports_direct_unmap()) {
/*
* Determine core local address of the region, where necessary.
* If we can't retrieve it, it is not possible to unmap on kernels
* that do not support direct unmap functionality, therefore return in that
* case.
* Otherwise calling flush with core_local address zero on kernels that
* unmap indirectly via core's address space can lead to illegitime unmaps
* of core memory (reference issue #3082)
*/
Address_space::Core_local_addr core_local { _core_local_addr(region) };
if (core_local.value)
platform_specific().core_pd().flush(0, region.size(), core_local);
} else {
/*
* Unmap this memory region from all region maps referencing it.
*/
_unmap_region(region.base(), region.size());
}
}
void Region_map_component::add_client(Rm_client &rm_client)
{
Lock::Guard lock_guard(_lock);
_clients.insert(&rm_client);
}
void Region_map_component::remove_client(Rm_client &rm_client)
{
Lock::Guard lock_guard(_lock);
_clients.remove(&rm_client);
rm_client.dissolve_from_faulting_region_map(*this);
}
void Region_map_component::fault(Rm_faulter &faulter, addr_t pf_addr,
Region_map::State::Fault_type pf_type)
{
/* remember fault state in faulting thread */
faulter.fault(*this, Region_map::State(pf_type, pf_addr));
/* enqueue faulter */
_faulters.enqueue(faulter);
/* issue fault signal */
_fault_notifier.submit();
}
void Region_map_component::discard_faulter(Rm_faulter &faulter, bool do_lock)
{
if (do_lock) {
Lock::Guard lock_guard(_lock);
_faulters.remove(faulter);
} else
_faulters.remove(faulter);
}
void Region_map_component::fault_handler(Signal_context_capability handler)
{
_fault_notifier.context(handler);
}
Region_map::State Region_map_component::state()
{
/* serialize access */
Lock::Guard lock_guard(_lock);
/* return ready state if there are not current faulters */
Region_map::State result;
/* otherwise return fault information regarding the first faulter */
_faulters.head([&] (Rm_faulter &faulter) {
result = faulter.fault_state(); });
return result;
}
static Dataspace_capability
_type_deduction_helper(Dataspace_capability cap) { return cap; }
Region_map_component::Region_map_component(Rpc_entrypoint &ep,
Allocator &md_alloc,
Pager_entrypoint &pager_ep,
addr_t vm_start,
size_t vm_size,
Session::Diag diag)
:
_diag(diag), _ds_ep(ep), _thread_ep(ep), _session_ep(ep),
_md_alloc(md_alloc),
_map(&_md_alloc), _pager_ep(pager_ep),
_ds(align_addr(vm_size, get_page_size_log2())),
_ds_cap(_type_deduction_helper(_ds_ep.manage(&_ds)))
{
/* configure managed VM area */
_map.add_range(vm_start, align_addr(vm_size, get_page_size_log2()));
Capability<Region_map> cap = ep.manage(this);
_ds.sub_rm(cap);
}
Region_map_component::~Region_map_component()
{
lock_for_destruction();
/*
* Normally, detaching ds from all regions maps is done in the
* destructor of the dataspace. But we do it here explicitely
* so that the regions refering this ds can retrieve it via
* there capabilities before it gets dissolved in the next step.
*/
_ds.detach_from_rm_sessions();
_ds_ep.dissolve(this);
/* dissolve all clients from pager entrypoint */
Rm_client *cl = nullptr;
do {
Cpu_session_capability cpu_session_cap;
Thread_capability thread_cap;
{
Lock::Guard lock_guard(_lock);
cl = _clients.first();
if (!cl) break;
cl->dissolve_from_faulting_region_map(*this);
cpu_session_cap = cl->cpu_session_cap();
thread_cap = cl->thread_cap();
_clients.remove(cl);
}
/* destroy thread */
auto lambda = [&] (Cpu_session_component *cpu_session) {
if (cpu_session)
cpu_session->kill_thread(thread_cap);
};
_thread_ep.apply(cpu_session_cap, lambda);
} while (cl);
/* detach all regions */
while (true) {
addr_t out_addr = 0;
{
Lock::Guard lock_guard(_lock);
if (!_map.any_block_addr(&out_addr))
break;
}
detach(out_addr);
}
/* revoke dataspace representation */
_ds_ep.dissolve(&_ds);
}