genode/repos/base-fiasco/src/core/platform.cc

536 lines
14 KiB
C++
Raw Normal View History

2011-12-22 16:19:25 +01:00
/*
* \brief Fiasco platform interface implementation
* \author Christian Helmuth
* \date 2006-04-11
*/
/*
2013-01-10 21:44:47 +01:00
* Copyright (C) 2006-2013 Genode Labs GmbH
2011-12-22 16:19:25 +01:00
*
* This file is part of the Genode OS framework, which is distributed
* under the terms of the GNU General Public License version 2.
*/
/* Genode includes */
#include <base/printf.h>
#include <base/allocator_avl.h>
#include <base/sleep.h>
#include <util/misc_math.h>
/* base-internal includes */
#include <base/internal/crt0.h>
#include <base/internal/fiasco_thread_helper.h>
#include <base/internal/stack_area.h>
2011-12-22 16:19:25 +01:00
/* core includes */
#include <core_parent.h>
#include <platform.h>
#include <platform_thread.h>
#include <platform_pd.h>
#include <util.h>
#include <multiboot.h>
/* Fiasco includes */
namespace Fiasco {
#include <l4/sys/types.h>
#include <l4/sys/syscalls.h>
#include <l4/sys/ipc.h>
#include <l4/sys/kernel.h>
#include <l4/sys/kip.h>
#include <l4/sigma0/sigma0.h>
}
using namespace Genode;
static const bool verbose = true;
static const bool verbose_core_pf = false;
static const bool verbose_region_alloc = false;
/***********************************
** Core address space management **
***********************************/
static Synced_range_allocator<Allocator_avl> &_core_address_ranges()
2011-12-22 16:19:25 +01:00
{
static Synced_range_allocator<Allocator_avl> _core_address_ranges(nullptr);
2011-12-22 16:19:25 +01:00
return _core_address_ranges;
}
enum { PAGER_STACK_ELEMENTS = 1024 };
static unsigned long _core_pager_stack[PAGER_STACK_ELEMENTS];
static unsigned _core_pager_arg;
/**
* Core pager "service loop"
*/
static void _core_pager_loop()
{
unsigned pd_id = _core_pager_arg;
using namespace Fiasco;
l4_threadid_t t;
l4_umword_t dw0, dw1;
l4_msgdope_t r;
bool send_reply = false;
while (1) {
if (send_reply)
/* unblock faulter and wait for next pagefault */
l4_ipc_reply_and_wait(t, L4_IPC_SHORT_MSG, 0, 0,
&t, L4_IPC_SHORT_MSG, &dw0, &dw1,
L4_IPC_NEVER, &r);
else
l4_ipc_wait(&t, L4_IPC_SHORT_MSG, &dw0, &dw1, L4_IPC_NEVER, &r);
/* ignore messages from non-core pds */
if (t.id.task != pd_id) break;
/* detect local map request */
if (dw1 == 0) {
l4_msgdope_t ipc_result;
l4_ipc_send(t, L4_IPC_SHORT_FPAGE, 0, dw0,
L4_IPC_SEND_TIMEOUT_0, &ipc_result);
send_reply = false;
continue;
}
bool rw = dw0 & 2;
addr_t pfa = dw0 & ~2;
if (pfa < L4_PAGESIZE) {
/* NULL pointer access */
PERR("Possible null pointer %s in %x.%02x at %lx IP %lx",
rw ? "WRITE" : "READ", (int)t.id.task, (int)t.id.lthread, pfa, dw1);
/* do not unblock faulter */
send_reply = false;
continue;
} else if (!_core_address_ranges().valid_addr(pfa)) {
/* page-fault address is not in RAM */
PERR("%s access outside of RAM in %x.%02x at %lx IP %lx",
rw ? "WRITE" : "READ", (int)t.id.task, (int)t.id.lthread, pfa, dw1);
/* do not unblock faulter */
send_reply = false;
continue;
} else if (verbose_core_pf)
PDBG("pfa=%lx ip=%lx thread %x.%02x", pfa, dw1, (int)t.id.task, (int)t.id.lthread);
/* my pf handler is sigma0 - just touch the appropriate page */
if (rw)
touch_rw((void *)pfa, 1);
else
touch_ro((void *)pfa, 1);
send_reply = true;
}
}
Platform::Sigma0::Sigma0() : Pager_object(0, Affinity::Location())
2011-12-22 16:19:25 +01:00
{
cap(reinterpret_cap_cast<Cpu_thread>(Native_capability(Fiasco::sigma0_threadid, 0)));
}
Platform::Sigma0 *Platform::sigma0()
{
static Sigma0 _sigma0;
return &_sigma0;
}
Platform::Core_pager::Core_pager(Platform_pd *core_pd)
:
thread API & CPU session: accounting of CPU quota In the init configuration one can configure the donation of CPU time via 'resource' tags that have the attribute 'name' set to "CPU" and the attribute 'quantum' set to the percentage of CPU quota that init shall donate. The pattern is the same as when donating RAM quota. ! <start name="test"> ! <resource name="CPU" quantum="75"/> ! </start> This would cause init to try donating 75% of its CPU quota to the child "test". Init and core do not preserve CPU quota for their own requirements by default as it is done with RAM quota. The CPU quota that a process owns can be applied through the thread constructor. The constructor has been enhanced by an argument that indicates the percentage of the programs CPU quota that shall be granted to the new thread. So 'Thread(33, "test")' would cause the backing CPU session to try to grant 33% of the programs CPU quota to the thread "test". By now, the CPU quota of a thread can't be altered after construction. Constructing a thread with CPU quota 0 doesn't mean the thread gets never scheduled but that the thread has no guaranty to receive CPU time. Such threads have to live with excess CPU time. Threads that already existed in the official repositories of Genode were adapted in the way that they receive a quota of 0. This commit also provides a run test 'cpu_quota' in base-hw (the only kernel that applies the CPU-quota scheme currently). The test basically runs three threads with different physical CPU quota. The threads simply count for 30 seconds each and the test then checks wether the counter values relate to the CPU-quota distribution. fix #1275
2014-10-16 11:15:46 +02:00
Platform_thread(0, "core.pager"), Pager_object(0, Affinity::Location())
2011-12-22 16:19:25 +01:00
{
Platform_thread::pager(sigma0());
core_pd->bind_thread(this);
cap(Native_capability(native_thread_id(), 0));
/* pager needs to know core's pd ID */
_core_pager_arg = core_pd->pd_id();
/* stack begins at the top end of the '_core_pager_stack' array */
void *sp = (void *)&_core_pager_stack[PAGER_STACK_ELEMENTS - 1];
start((void *)_core_pager_loop, sp);
using namespace Fiasco;
/* pager0 receives pagefaults from me - for NULL pointer detection */
l4_umword_t d;
l4_threadid_t preempter = L4_INVALID_ID;
l4_threadid_t pager = native_thread_id();
l4_thread_ex_regs(l4_myself(), ~0UL, ~0UL, &preempter, &pager, &d, &d, &d);
}
Platform::Core_pager *Platform::core_pager()
{
static Core_pager _core_pager(core_pd());
return &_core_pager;
}
/***********************************
** Helper for L4 region handling **
***********************************/
struct Region
{
addr_t start;
addr_t end;
Region() : start(0), end(0) { }
Region(addr_t s, addr_t e) : start(s), end(e) { }
/**
* Returns true if the specified range intersects with the region
*/
bool intersects(addr_t base, size_t size) const
{
return (((base + size) > start) && (base < end));
}
2011-12-22 16:19:25 +01:00
};
/**
* Log region
*/
static inline void print_region(Region r)
{
printf("[%08lx,%08lx) %08lx", r.start, r.end, r.end - r.start);
}
/**
* Add region to allocator
*/
static inline void add_region(Region r, Range_allocator &alloc)
{
if (verbose_region_alloc) {
printf("%p add: ", &alloc); print_region(r); printf("\n");
}
/* adjust region */
addr_t start = trunc_page(r.start);
addr_t end = round_page(r.end);
alloc.add_range(start, end - start);
}
/**
* Remove region from allocator
*/
static inline void remove_region(Region r, Range_allocator &alloc)
{
if (verbose_region_alloc) {
printf("%p remove: ", &alloc); print_region(r); printf("\n");
}
/* adjust region */
addr_t start = trunc_page(r.start);
addr_t end = round_page(r.end);
alloc.remove_range(start, end - start);
}
/**
* Request any RAM page from Sigma0
*/
static inline int sigma0_req_region(addr_t *addr, unsigned log2size)
{
using namespace Fiasco;
/* XXX sigma0 always maps pages RW */
l4_umword_t req_fpage = l4_fpage(0, log2size, 0, 0).fpage;
void* rcv_window = L4_IPC_MAPMSG(0, L4_WHOLE_ADDRESS_SPACE);
addr_t base;
l4_fpage_t rcv_fpage;
l4_msgdope_t result;
l4_msgtag_t tag;
int err = l4_ipc_call_tag(Fiasco::sigma0_threadid,
L4_IPC_SHORT_MSG, SIGMA0_REQ_FPAGE_ANY, req_fpage,
l4_msgtag(L4_MSGTAG_SIGMA0, 0, 0, 0),
rcv_window, &base, (l4_umword_t *)&rcv_fpage,
L4_IPC_NEVER, &result, &tag);
int ret = (err || !l4_ipc_fpage_received(result));
if (!ret) touch_rw((void *)addr, 1);
*addr = base;
return ret;
}
void Platform::_setup_mem_alloc()
{
/*
* Completely map program image by touching all pages read-only to
* prevent sigma0 from handing out those page as anonymous memory.
*/
volatile const char *beg, *end;
beg = (const char *)(((unsigned)&_prog_img_beg) & L4_PAGEMASK);
end = (const char *)&_prog_img_end;
for ( ; beg < end; beg += L4_PAGESIZE) (void)(*beg);
/* request pages of known page size starting with largest */
size_t log2_sizes[] = { L4_LOG2_SUPERPAGESIZE, L4_LOG2_PAGESIZE };
for (unsigned i = 0; i < sizeof(log2_sizes)/sizeof(*log2_sizes); ++i) {
size_t log2_size = log2_sizes[i];
size_t size = 1 << log2_size;
int err = 0;
addr_t addr;
Region region;
/* request any page of current size from sigma0 */
do {
err = sigma0_req_region(&addr, log2_size);
if (!err) {
/* XXX do not allocate page0 */
if (addr == 0) {
Fiasco::l4_fpage_unmap(Fiasco::l4_fpage(0, log2_size, 0, 0),
L4_FP_FLUSH_PAGE | L4_FP_ALL_SPACES);
continue;
}
region.start = addr; region.end = addr + size;
if (!region.intersects(stack_area_virtual_base(),
stack_area_virtual_size())) {
add_region(region, _ram_alloc);
add_region(region, _core_address_ranges());
}
2011-12-22 16:19:25 +01:00
remove_region(region, _io_mem_alloc);
remove_region(region, _region_alloc);
}
} while (!err);
}
}
void Platform::_setup_irq_alloc() {
_irq_alloc.add_range(0, 0x10); }
static Fiasco::l4_kernel_info_t *get_kip()
2011-12-22 16:19:25 +01:00
{
using namespace Fiasco;
int err;
/* region allocator is not setup yet */
/* map KIP one-to-one */
void *fpage = L4_IPC_MAPMSG(0, L4_WHOLE_ADDRESS_SPACE);
l4_umword_t dw0, dw1;
l4_msgdope_t r;
l4_msgtag_t tag;
err = l4_ipc_call_tag(Fiasco::sigma0_threadid,
L4_IPC_SHORT_MSG, SIGMA0_REQ_KIP, 0,
l4_msgtag(L4_MSGTAG_SIGMA0, 0, 0, 0),
fpage, &dw0, &dw1,
L4_IPC_NEVER, &r, &tag);
bool amok = false;
if (err) {
printf("IPC error %d\n", err);
amok = true;
}
if (!l4_ipc_fpage_received(r)) {
printf("No fpage received\n");
amok = true;
}
if (amok)
panic("kip mapping failed");
/* store mapping base from received mapping */
l4_kernel_info_t *kip = (l4_kernel_info_t *)dw0;
if (kip->magic != L4_KERNEL_INFO_MAGIC)
panic("Sigma0 mapped something but not the KIP");
if (verbose) {
printf("\n");
printf("KIP @ %p\n", kip);
printf(" magic: %08x\n", kip->magic);
printf(" version: %08x\n", kip->version);
printf(" sigma0 "); printf(" esp: %08lx eip: %08lx\n", kip->sigma0_esp, kip->sigma0_eip);
printf(" sigma1 "); printf(" esp: %08lx eip: %08lx\n", kip->sigma1_esp, kip->sigma1_eip);
printf(" root "); printf(" esp: %08lx eip: %08lx\n", kip->root_esp, kip->root_eip);
}
return kip;
}
void Platform::_setup_basics()
{
using namespace Fiasco;
l4_kernel_info_t * kip = get_kip();
2011-12-22 16:19:25 +01:00
/* add KIP as ROM module */
_kip_rom = Rom_module((addr_t)kip, L4_PAGESIZE, "l4v2_kip");
_rom_fs.insert(&_kip_rom);
/* update multi-boot info pointer from KIP */
addr_t mb_info_addr = kip->user_ptr;
if (verbose) printf("MBI @ 0x%lx\n", mb_info_addr);
2011-12-22 16:19:25 +01:00
/* parse memory descriptors - look for virtual memory configuration */
/* XXX we support only one VM region (here and also inside RM) */
using L4::Kip::Mem_desc;
_vm_start = 0; _vm_size = 0;
Mem_desc *desc = Mem_desc::first(kip);
for (unsigned i = 0; i < Mem_desc::count(kip); ++i)
if (desc[i].is_virtual()) {
_vm_start = round_page(desc[i].start());
_vm_size = trunc_page(desc[i].end() - _vm_start + 1);
break;
}
if (_vm_size == 0)
panic("Virtual memory configuration not found");
/* configure applicable address space but never use page0 */
_vm_size = _vm_start == 0 ? _vm_size - L4_PAGESIZE : _vm_size;
_vm_start = _vm_start == 0 ? L4_PAGESIZE : _vm_start;
_region_alloc.add_range(_vm_start, _vm_size);
/* preserve stack area in core's virtual address space */
_region_alloc.remove_range(stack_area_virtual_base(),
stack_area_virtual_size());
2011-12-22 16:19:25 +01:00
/* I/O memory could be the whole user address space */
/* FIXME if the kernel helps to find out max address - use info here */
_io_mem_alloc.add_range(0, ~0);
/* remove KIP and MBI area from region and IO_MEM allocator */
remove_region(Region((addr_t)kip, (addr_t)kip + L4_PAGESIZE), _region_alloc);
remove_region(Region((addr_t)kip, (addr_t)kip + L4_PAGESIZE), _io_mem_alloc);
remove_region(Region(mb_info_addr, mb_info_addr + _mb_info.size()), _region_alloc);
remove_region(Region(mb_info_addr, mb_info_addr + _mb_info.size()), _io_mem_alloc);
2011-12-22 16:19:25 +01:00
/* remove core program image memory from region and IO_MEM allocator */
addr_t img_start = (addr_t) &_prog_img_beg;
addr_t img_end = (addr_t) &_prog_img_end;
remove_region(Region(img_start, img_end), _region_alloc);
remove_region(Region(img_start, img_end), _io_mem_alloc);
/* image is accessible by core */
add_region(Region(img_start, img_end), _core_address_ranges());
}
void Platform::_setup_rom()
{
Rom_module rom;
for (unsigned i = FIRST_ROM; i < _mb_info.num_modules(); i++) {
if (!(rom = _mb_info.get_module(i)).valid()) continue;
Rom_module *new_rom = new(core_mem_alloc()) Rom_module(rom);
_rom_fs.insert(new_rom);
if (verbose)
printf(" mod[%d] [%p,%p) %s\n", i,
(void *)new_rom->addr(), ((char *)new_rom->addr()) + new_rom->size(),
new_rom->name());
/* zero remainder of last ROM page */
size_t count = L4_PAGESIZE - rom.size() % L4_PAGESIZE;
if (count != L4_PAGESIZE)
memset(reinterpret_cast<void *>(rom.addr() + rom.size()), 0, count);
/* remove ROM area from region and IO_MEM allocator */
remove_region(Region(new_rom->addr(), new_rom->addr() + new_rom->size()), _region_alloc);
remove_region(Region(new_rom->addr(), new_rom->addr() + new_rom->size()), _io_mem_alloc);
/* add area to core-accessible ranges */
add_region(Region(new_rom->addr(), new_rom->addr() + new_rom->size()), _core_address_ranges());
}
}
Platform::Platform() :
_ram_alloc(nullptr), _io_mem_alloc(core_mem_alloc()),
2011-12-22 16:19:25 +01:00
_io_port_alloc(core_mem_alloc()), _irq_alloc(core_mem_alloc()),
_region_alloc(core_mem_alloc()),
_mb_info(get_kip()->user_ptr, true)
2011-12-22 16:19:25 +01:00
{
/*
* We must be single-threaded at this stage and so this is safe.
*/
static bool initialized = 0;
if (initialized) panic("Platform constructed twice!");
initialized = true;
_setup_basics();
_setup_mem_alloc();
_setup_io_port_alloc();
_setup_irq_alloc();
_setup_rom();
if (verbose) {
printf(":ram_alloc: "); _ram_alloc()->dump_addr_tree();
printf(":region_alloc: "); _region_alloc()->dump_addr_tree();
printf(":io_mem: "); _io_mem_alloc()->dump_addr_tree();
printf(":io_port: "); _io_port_alloc()->dump_addr_tree();
printf(":irq: "); _irq_alloc()->dump_addr_tree();
2011-12-22 16:19:25 +01:00
printf(":rom_fs: "); _rom_fs.print_fs();
printf(":core ranges: "); _core_address_ranges()()->dump_addr_tree();
2011-12-22 16:19:25 +01:00
}
Fiasco::l4_threadid_t myself = Fiasco::l4_myself();
Platform_pd::init();
/* setup pd object for core pd */
_core_label[0] = 0;
_core_pd = new(core_mem_alloc()) Platform_pd(nullptr, _core_label,
myself.id.task, false);
2011-12-22 16:19:25 +01:00
/*
* We setup the thread object for thread0 in core pd using a special
* interface that allows us to specify the lthread number.
*/
thread API & CPU session: accounting of CPU quota In the init configuration one can configure the donation of CPU time via 'resource' tags that have the attribute 'name' set to "CPU" and the attribute 'quantum' set to the percentage of CPU quota that init shall donate. The pattern is the same as when donating RAM quota. ! <start name="test"> ! <resource name="CPU" quantum="75"/> ! </start> This would cause init to try donating 75% of its CPU quota to the child "test". Init and core do not preserve CPU quota for their own requirements by default as it is done with RAM quota. The CPU quota that a process owns can be applied through the thread constructor. The constructor has been enhanced by an argument that indicates the percentage of the programs CPU quota that shall be granted to the new thread. So 'Thread(33, "test")' would cause the backing CPU session to try to grant 33% of the programs CPU quota to the thread "test". By now, the CPU quota of a thread can't be altered after construction. Constructing a thread with CPU quota 0 doesn't mean the thread gets never scheduled but that the thread has no guaranty to receive CPU time. Such threads have to live with excess CPU time. Threads that already existed in the official repositories of Genode were adapted in the way that they receive a quota of 0. This commit also provides a run test 'cpu_quota' in base-hw (the only kernel that applies the CPU-quota scheme currently). The test basically runs three threads with different physical CPU quota. The threads simply count for 30 seconds each and the test then checks wether the counter values relate to the CPU-quota distribution. fix #1275
2014-10-16 11:15:46 +02:00
Platform_thread *core_thread = new(core_mem_alloc())
Platform_thread(0, "core.main", myself.id.lthread);
2011-12-22 16:19:25 +01:00
core_thread->pager(sigma0());
_core_pd->bind_thread(core_thread);
/* we never call _core_thread.start(), so set name directly */
Fiasco::fiasco_register_thread_name(core_thread->native_thread_id(), core_thread->name());
}
/********************************
** Generic platform interface **
********************************/
void Platform::wait_for_exit()
{
/*
* On Fiasco, Core never exits. So let us sleep forever.
*/
sleep_forever();
}
void Core_parent::exit(int exit_value) { }