Malloc / free ansd rump_fs ->Slab-backend exhausted
w_schmidt at ...181...
w_schmidt at ...181...
Mon Aug 18 20:09:04 CEST 2014
Hi,
>> this seems to be happen only if rump_fs has a larger quota – the file
>> can be larger if rump_fs has a smaller quota. (memory tests of the
>> system have shown no errors so RAM should be okay)
>this is indeed very strange. I hope you have a Genode version where
>there is no Slab-Backend allocator, have a look at
'>include/util/allocator_fap.h" in the dde_rump repository, there should
>only be a PERR message with the slab warning left.
The file looks like below,
Best regards,
wOlfgang
/**
* \brief Fast allocator for porting
* \author Sebastian Sumpf
* \date 2013-06-12
*/
/*
* Copyright (C) 2013-2014 Genode Labs GmbH
*
* This file is part of the Genode OS framework, which is distributed
* under the terms of the GNU General Public License version 2.
*/
#ifndef _INCLUDE__UTIL__ALLOCATOR_FAP_H_
#define _INCLUDE__UTIL__ALLOCATOR_FAP_H_
#include <base/allocator_avl.h>
#include <dataspace/client.h>
#include <rm_session/connection.h>
namespace Allocator {
template <unsigned VM_SIZE, typename POLICY> class Backend_alloc;
template <unsigned VM_SIZE, typename POLICY> class Fap;
}
namespace Allocator {
using namespace Genode;
struct Default_allocator_policy
{
static int block() { return 0; }
static void unblock(int) { }
};
template <typename POLICY>
struct Policy_guard
{
int val;
Policy_guard() { val = POLICY::block(); }
~Policy_guard() { POLICY::unblock(val); }
};
/**
* Back-end allocator for Genode's slab allocator
*/
template <unsigned VM_SIZE, typename POLICY = Default_allocator_policy>
class Backend_alloc : public Genode::Allocator,
public Genode::Rm_connection
{
private:
enum {
BLOCK_SIZE = 1024 * 1024, /* 1 MB */
ELEMENTS = VM_SIZE / BLOCK_SIZE, /* MAX number of
dataspaces in VM */
};
typedef Genode::addr_t addr_t;
typedef Genode::Ram_dataspace_capability
Ram_dataspace_capability;
typedef Genode::Allocator_avl Allocator_avl;
addr_t _base; /* virt. base
address */
Cache_attribute _cached; /* non-/cached RAM
*/
Ram_dataspace_capability _ds_cap[ELEMENTS]; /* dataspaces to
put in VM */
addr_t _ds_phys[ELEMENTS]; /* physical bases
of dataspaces */
int _index = 0; /* current index in
ds_cap */
Allocator_avl _range; /* manage
allocations */
bool _quota_exceeded = false;
bool _alloc_block()
{
if (_quota_exceeded)
return false;
if (_index == ELEMENTS) {
PERR("Slab-backend exhausted!");
return false;
}
Policy_guard<POLICY> guard;
try {
_ds_cap[_index] =
Genode::env()->ram_session()->alloc(BLOCK_SIZE, _cached);
/* attach at index * BLOCK_SIZE */
Rm_connection::attach_at(_ds_cap[_index], _index *
BLOCK_SIZE, BLOCK_SIZE, 0);
/* lookup phys. address */
_ds_phys[_index] =
Genode::Dataspace_client(_ds_cap[_index]).phys_addr();
} catch (Genode::Ram_session::Quota_exceeded) {
PWRN("Backend allocator exhausted");
_quota_exceeded = true;
return false;
} catch (Genode::Rm_session::Attach_failed) {
PWRN("Backend VM region exhausted");
_quota_exceeded = true;
return false;
}
/* return base + offset in VM area */
addr_t block_base = _base + (_index * BLOCK_SIZE);
++_index;
_range.add_range(block_base, BLOCK_SIZE);
return true;
}
public:
Backend_alloc(Cache_attribute cached)
: Rm_connection(0, VM_SIZE), _cached(cached),
_range(Genode::env()->heap())
{
/* reserver attach us, anywere */
_base = Genode::env()->rm_session()->attach(dataspace());
}
/**
* Allocate
*/
bool alloc(size_t size, void **out_addr)
{
bool done = _range.alloc(size, out_addr);
if (done)
return done;
done = _alloc_block();
if (!done)
return false;
return _range.alloc(size, out_addr);
}
void *alloc_aligned(size_t size, int align = 0)
{
void *addr;
if (!_range.alloc_aligned(size, &addr, align).is_error())
return addr;
if (!_alloc_block())
return 0;
if (_range.alloc_aligned(size, &addr, align).is_error()) {
PERR("Backend allocator: Unable to allocate memory
(size: %zu align: %d:)",
size, align);
return 0;
}
return addr;
}
void free(void *addr, size_t size) { _range.free(addr,
size); }
size_t overhead(size_t size) { return 0; }
bool need_size_for_free() const override { return false; }
/**
* Return phys address for given virtual addr.
*/
addr_t phys_addr(addr_t addr)
{
if (addr < _base || addr >= (_base + VM_SIZE))
return ~0UL;
int index = (addr - _base) / BLOCK_SIZE;
/* physical base of dataspace */
addr_t phys = _ds_phys[index];
if (!phys)
return ~0UL;
/* add offset */
phys += (addr - _base - (index * BLOCK_SIZE));
return phys;
}
bool inside(addr_t addr) const { return (addr >= _base) && (addr
< (_base + VM_SIZE)); }
};
/**
* Interface
*/
template <unsigned VM_SIZE, typename POLICY = Default_allocator_policy>
class Fap
{
private:
typedef Allocator::Backend_alloc<VM_SIZE, POLICY> Backend_alloc;
Backend_alloc _back_allocator;
public:
Fap(bool cached)
: _back_allocator(cached ? CACHED : UNCACHED) { }
void *alloc(size_t size, int align = 0)
{
return _back_allocator.alloc_aligned(size, align);
}
void free(void *addr, size_t size)
{
_back_allocator.free(addr, size);
}
addr_t phys_addr(void *addr)
{
return _back_allocator.phys_addr((addr_t)addr);
}
};
} /* namespace Allocator */
#endif /* _INCLUDE__UTIL__ALLOCATOR_FAP_H_ */
More information about the users
mailing list