/*===========================================================================*
* req_newdriver *
*===========================================================================*/
PUBLIC int req_newdriver(
endpoint_t fs_e,
dev_t dev,
endpoint_t driver_e
)
{
/* Note: this is the only request function that doesn't use the
* fs_sendrec internal routine, since we want to avoid the dead
* driver recovery mechanism here. This function is actually called
* during the recovery.
*/
message m;
int r;
/* Fill in request message */
m.m_type = REQ_NEW_DRIVER;
m.REQ_DEV = dev;
m.REQ_DRIVER_E = driver_e;
/* Issue request */
if((r = sendrec(fs_e, &m)) != OK) {
printf("%s:%d VFS req_newdriver: error sending message %d to %d\n",
__FILE__, __LINE__, r, fs_e);
util_stacktrace();
return(r);
}
return(OK);
}
/*===========================================================================*
* put_vnode *
*===========================================================================*/
PUBLIC void put_vnode(struct vnode *vp)
{
/* Decrease vnode's usage counter and decrease inode's usage counter in the
* corresponding FS process. Decreasing the fs_count each time we decrease the
* ref count would lead to poor performance. Instead, only decrease fs_count
* when the ref count hits zero. However, this could lead to fs_count to wrap.
* To prevent this, we drop the counter to 1 when the counter hits 256.
* We maintain fs_count as a sanity check to make sure VFS and the FS are in
* sync.
*/
int r, lock_vp;
ASSERTVP(vp);
/* Lock vnode. It's quite possible this thread already has a lock on this
* vnode. That's no problem, because the reference counter will not decrease
* to zero in that case. However, if the counter does decrease to zero *and*
* is already locked, we have a consistency problem somewhere. */
lock_vp = lock_vnode(vp, VNODE_OPCL);
if (vp->v_ref_count > 1) {
/* Decrease counter */
vp->v_ref_count--;
if (vp->v_fs_count > 256)
vnode_clean_refs(vp);
if (lock_vp != EBUSY) unlock_vnode(vp);
return;
}
/* If we already had a lock, there is a consistency problem */
assert(lock_vp != EBUSY);
tll_upgrade(&vp->v_lock); /* Make sure nobody else accesses this vnode */
/* A vnode that's not in use can't be put back. */
if (vp->v_ref_count <= 0)
panic("put_vnode failed: bad v_ref_count %d\n", vp->v_ref_count);
/* fs_count should indicate that the file is in use. */
if (vp->v_fs_count <= 0)
panic("put_vnode failed: bad v_fs_count %d\n", vp->v_fs_count);
/* Tell FS we don't need this inode to be open anymore. */
r = req_putnode(vp->v_fs_e, vp->v_inode_nr, vp->v_fs_count);
if (r != OK) {
printf("VFS: putnode failed: %d\n", r);
util_stacktrace();
}
/* This inode could've been mapped. If so, tell mapped FS to close it as
* well. If mapped onto same FS, this putnode is not needed. */
if (vp->v_mapfs_e != NONE && vp->v_mapfs_e != vp->v_fs_e)
req_putnode(vp->v_mapfs_e, vp->v_mapinode_nr, vp->v_mapfs_count);
vp->v_fs_count = 0;
vp->v_ref_count = 0;
vp->v_mapfs_count = 0;
unlock_vnode(vp);
}
static int getsrc(struct vir_region *region,
struct vmproc **vmp, struct vir_region **r)
{
int srcproc;
if(region->def_memtype != &mem_type_shared) {
printf("shared region hasn't shared type but %s.\n",
region->def_memtype->name);
return EINVAL;
}
if(!region->param.shared.ep || !region->param.shared.vaddr) {
printf("shared region has not defined source region.\n");
util_stacktrace();
return EINVAL;
}
if(vm_isokendpt((endpoint_t) region->param.shared.ep, &srcproc) != OK) {
printf("VM: shared memory with missing source process.\n");
util_stacktrace();
return EINVAL;
}
*vmp = &vmproc[srcproc];
if(!(*r=map_lookup(*vmp, region->param.shared.vaddr, NULL))) {
printf("VM: shared memory with missing vaddr 0x%lx.\n",
region->param.shared.vaddr);
return EINVAL;
}
if((*r)->def_memtype != &mem_type_anon) {
printf("source region hasn't anon type but %s.\n",
(*r)->def_memtype->name);
return EINVAL;
}
if(region->param.shared.id != (*r)->id) {
printf("source region has no matching id\n");
return EINVAL;
}
return OK;
}
/*===========================================================================*
* reply *
*===========================================================================*/
void reply(endpoint_t whom, int result)
{
/* Send a reply to a user process. If the send fails, just ignore it. */
int r;
m_out.reply_type = result;
r = sendnb(whom, &m_out);
if (r != OK) {
printf("VFS: %d couldn't send reply %d to %d: %d\n", mthread_self(),
result, whom, r);
util_stacktrace();
}
}
/*===========================================================================*
* panic *
*===========================================================================*/
void panic(const char *fmt, ...)
{
/* Something awful has happened. Panics are caused when an internal
* inconsistency is detected, e.g., a programming error or illegal
* value of a defined constant.
*/
endpoint_t me = NONE;
char name[20];
int priv_flags;
int init_flags;
void (*suicide)(void);
va_list args;
if(sys_whoami(&me, name, sizeof(name), &priv_flags, &init_flags) == OK && me != NONE)
printf("%s(%d): panic: ", name, me);
else
printf("(sys_whoami failed): panic: ");
if(fmt) {
va_start(args, fmt);
vprintf(fmt, args);
va_end(args);
} else {
printf("no message\n");
}
printf("\n");
printf("syslib:panic.c: stacktrace: ");
util_stacktrace();
panic_hook();
/* Try exit */
_exit(1);
/* Try to signal ourself */
abort();
/* If exiting nicely through PM fails for some reason, try to
* commit suicide. E.g., message to PM might fail due to deadlock.
*/
suicide = (void (*)(void)) -1;
suicide();
/* If committing suicide fails for some reason, hang. */
for(;;) { }
}
/*===========================================================================*
* minix_panic *
*===========================================================================*/
PUBLIC void minix_panic(char *mess,int nr)
{
/* The system has run aground of a fatal kernel error. Terminate execution. */
if (minix_panicing++) {
arch_monitor();
}
if (mess != NULL) {
kprintf("kernel panic: %s", mess);
if(nr != NO_NUM)
kprintf(" %d", nr);
kprintf("\n");
}
kprintf("kernel: ");
util_stacktrace();
/* Abort MINIX. */
minix_shutdown(NULL);
}
void kernel_panic(char *mess, int nr)
{
/* But what if we are already in panic? */
kernel_in_panic++;
if (mess != NULL) {
printk("kernel panic: %s", mess);
if(nr != NO_NUM)
printk(" %d", nr);
printk("\n");
}
printk("kernel: ");
util_stacktrace();
/* Abort Nucleos. */
nucleos_shutdown(RBT_HALT);
}
/*===========================================================================*
* panic *
*===========================================================================*/
PUBLIC void panic(const char *fmt, ...)
{
va_list arg;
/* The system has run aground of a fatal kernel error. Terminate execution. */
if (minix_panicing == ARE_PANICING) {
arch_monitor();
}
minix_panicing = ARE_PANICING;
if (fmt != NULL) {
printf("kernel panic: ");
va_start(arg, fmt);
vprintf(fmt, arg);
printf("\n");
}
printf("kernel: ");
util_stacktrace();
/* Abort MINIX. */
minix_shutdown(NULL);
}
/*===========================================================================*
* sendmsg *
*===========================================================================*/
static int sendmsg(struct vmnt *vmp, endpoint_t dst, struct worker_thread *wp)
{
/* This is the low level function that sends requests.
* Currently to FSes or VM.
*/
int r, transid;
if(vmp) vmp->m_comm.c_cur_reqs++; /* One more request awaiting a reply */
transid = wp->w_tid + VFS_TRANSID;
wp->w_sendrec->m_type = TRNS_ADD_ID(wp->w_sendrec->m_type, transid);
wp->w_task = dst;
if ((r = asynsend3(dst, wp->w_sendrec, AMF_NOREPLY)) != OK) {
printf("VFS: sendmsg: error sending message. "
"dest: %d req_nr: %d err: %d\n", dst,
wp->w_sendrec->m_type, r);
util_stacktrace();
return(r);
}
return(r);
}
/*===========================================================================*
* drv_sendrec *
*===========================================================================*/
int drv_sendrec(endpoint_t drv_e, message *reqmp)
{
int r;
struct dmap *dp;
/* For the CTTY_MAJOR case, we would actually have to lock the device
* entry being redirected to. However, the CTTY major only hosts a
* character device while this function is used only for block devices.
* Thus, we can simply deny the request immediately.
*/
if (drv_e == CTTY_ENDPT) {
printf("VFS: /dev/tty is not a block device!\n");
return EIO;
}
if ((dp = get_dmap(drv_e)) == NULL)
panic("driver endpoint %d invalid", drv_e);
lock_dmap(dp);
if (dp->dmap_servicing != INVALID_THREAD)
panic("driver locking inconsistency");
dp->dmap_servicing = self->w_tid;
self->w_task = drv_e;
self->w_drv_sendrec = reqmp;
if ((r = asynsend3(drv_e, self->w_drv_sendrec, AMF_NOREPLY)) == OK) {
/* Yield execution until we've received the reply */
worker_wait();
} else {
printf("VFS: drv_sendrec: error sending msg to driver %d: %d\n",
drv_e, r);
util_stacktrace();
}
dp->dmap_servicing = INVALID_THREAD;
self->w_task = NONE;
self->w_drv_sendrec = NULL;
unlock_dmap(dp);
return(OK);
}
void arch_proc_setcontext(struct proc *p, struct stackframe_s *state,
int isuser, int trap_style)
{
if(isuser) {
/* Restore user bits of psw from sc, maintain system bits
* from proc.
*/
state->psw = (state->psw & X86_FLAGS_USER) |
(p->p_reg.psw & ~X86_FLAGS_USER);
}
/* someone wants to totally re-initialize process state */
assert(sizeof(p->p_reg) == sizeof(*state));
memcpy(&p->p_reg, state, sizeof(*state));
/* further code is instructed to not touch the context
* any more
*/
p->p_misc_flags |= MF_CONTEXT_SET;
/* on x86 this requires returning using iret (KTS_INT)
* so that the full context is restored instead of relying on
* the userspace doing it (as it would do on SYSEXIT).
* as ESP and EIP are also reset, userspace won't try to
* restore bogus context after returning.
*
* if the process is not blocked, or the kernel will ignore
* our trap style, we needn't panic but things will probably
* not go well for the process (restored context will be ignored)
* and the situation should be debugged.
*/
if(!(p->p_rts_flags)) {
printf("WARNINIG: setting full context of runnable process\n");
print_proc(p);
util_stacktrace();
}
if(p->p_seg.p_kern_trap_style == KTS_NONE)
printf("WARNINIG: setting full context of out-of-kernel process\n");
p->p_seg.p_kern_trap_style = trap_style;
}
/*===========================================================================*
* usedpages_add *
*===========================================================================*/
int usedpages_add_f(phys_bytes addr, phys_bytes len, char *file, int line)
{
u32_t pagestart, pages;
if(!incheck)
return OK;
assert(!(addr % VM_PAGE_SIZE));
assert(!(len % VM_PAGE_SIZE));
assert(len > 0);
pagestart = addr / VM_PAGE_SIZE;
pages = len / VM_PAGE_SIZE;
while(pages > 0) {
phys_bytes thisaddr;
assert(pagestart > 0);
assert(pagestart < NUMBER_PHYSICAL_PAGES);
thisaddr = pagestart * VM_PAGE_SIZE;
assert(pagestart >= 0);
assert(pagestart < NUMBER_PHYSICAL_PAGES);
if(pagemap[pagestart].used) {
static int warnings = 0;
if(warnings++ < 100)
printf("%s:%d: usedpages_add: addr 0x%lx reused, first %s:%d\n",
file, line, thisaddr, pagemap[pagestart].file, pagemap[pagestart].line);
util_stacktrace();
return EFAULT;
}
pagemap[pagestart].used = 1;
pagemap[pagestart].file = file;
pagemap[pagestart].line = line;
pages--;
pagestart++;
}
return OK;
}
/*===========================================================================*
* panic *
*===========================================================================*/
PUBLIC void panic(const char *fmt, ...)
{
va_list arg;
/* The system has run aground of a fatal kernel error. Terminate execution. */
if (minix_panicing == ARE_PANICING) {
reset();
}
minix_panicing = ARE_PANICING;
if (fmt != NULL) {
printf("kernel panic: ");
va_start(arg, fmt);
vprintf(fmt, arg);
printf("\n");
}
printf("kernel on CPU %d: ", cpuid);
util_stacktrace();
printf("current process : ");
proc_stacktrace(get_cpulocal_var(proc_ptr));
/* Abort MINIX. */
minix_shutdown(NULL);
}
/*===========================================================================*
* vm_allocpage *
*===========================================================================*/
void *vm_allocpages(phys_bytes *phys, int reason, int pages)
{
/* Allocate a page for use by VM itself. */
phys_bytes newpage;
vir_bytes loc;
pt_t *pt;
int r;
static int level = 0;
void *ret;
u32_t mem_flags = 0;
pt = &vmprocess->vm_pt;
assert(reason >= 0 && reason < VMP_CATEGORIES);
assert(pages > 0);
level++;
assert(level >= 1);
assert(level <= 2);
if((level > 1) || !pt_init_done) {
void *s;
if(pages == 1) s=vm_getsparepage(phys);
else if(pages == 4) s=vm_getsparepagedir(phys);
else panic("%d pages", pages);
level--;
if(!s) {
util_stacktrace();
printf("VM: warning: out of spare pages\n");
}
if(!is_staticaddr(s)) vm_self_pages++;
return s;
}
#if defined(__arm__)
if (reason == VMP_PAGEDIR) {
mem_flags |= PAF_ALIGN16K;
}
#endif
/* VM does have a pagetable, so get a page and map it in there.
* Where in our virtual address space can we put it?
*/
loc = findhole(pages);
if(loc == NO_MEM) {
level--;
printf("VM: vm_allocpage: findhole failed\n");
return NULL;
}
/* Allocate page of memory for use by VM. As VM
* is trusted, we don't have to pre-clear it.
*/
if((newpage = alloc_mem(pages, mem_flags)) == NO_MEM) {
level--;
printf("VM: vm_allocpage: alloc_mem failed\n");
return NULL;
}
*phys = CLICK2ABS(newpage);
/* Map this page into our address space. */
if((r=pt_writemap(vmprocess, pt, loc, *phys, VM_PAGE_SIZE*pages,
ARCH_VM_PTE_PRESENT | ARCH_VM_PTE_USER | ARCH_VM_PTE_RW
#if defined(__arm__)
| ARM_VM_PTE_WB | ARM_VM_PTE_SHAREABLE
#endif
, 0)) != OK) {
free_mem(newpage, pages);
printf("vm_allocpage writemap failed\n");
level--;
return NULL;
}
if((r=sys_vmctl(SELF, VMCTL_FLUSHTLB, 0)) != OK) {
panic("VMCTL_FLUSHTLB failed: %d", r);
}
level--;
/* Return user-space-ready pointer to it. */
ret = (void *) loc;
vm_self_pages++;
return ret;
}
/*===========================================================================*
* lmfs_get_block_ino *
*===========================================================================*/
struct buf *lmfs_get_block_ino(dev_t dev, block_t block, int only_search,
ino_t ino, u64_t ino_off)
{
/* Check to see if the requested block is in the block cache. If so, return
* a pointer to it. If not, evict some other block and fetch it (unless
* 'only_search' is 1). All the blocks in the cache that are not in use
* are linked together in a chain, with 'front' pointing to the least recently
* used block and 'rear' to the most recently used block. If 'only_search' is
* 1, the block being requested will be overwritten in its entirety, so it is
* only necessary to see if it is in the cache; if it is not, any free buffer
* will do. It is not necessary to actually read the block in from disk.
* If 'only_search' is PREFETCH, the block need not be read from the disk,
* and the device is not to be marked on the block, so callers can tell if
* the block returned is valid.
* In addition to the LRU chain, there is also a hash chain to link together
* blocks whose block numbers end with the same bit strings, for fast lookup.
*/
int b;
static struct buf *bp;
u64_t dev_off = (u64_t) block * fs_block_size;
struct buf *prev_ptr;
assert(buf_hash);
assert(buf);
assert(nr_bufs > 0);
ASSERT(fs_block_size > 0);
assert(dev != NO_DEV);
if((ino_off % fs_block_size)) {
printf("cache: unaligned lmfs_get_block_ino ino_off %llu\n",
ino_off);
util_stacktrace();
}
/* Search the hash chain for (dev, block). */
b = BUFHASH(block);
bp = buf_hash[b];
while (bp != NULL) {
if (bp->lmfs_blocknr == block && bp->lmfs_dev == dev) {
if(bp->lmfs_flags & VMMC_EVICTED) {
/* We had it but VM evicted it; invalidate it. */
ASSERT(bp->lmfs_count == 0);
ASSERT(!(bp->lmfs_flags & VMMC_BLOCK_LOCKED));
ASSERT(!(bp->lmfs_flags & VMMC_DIRTY));
bp->lmfs_dev = NO_DEV;
bp->lmfs_bytes = 0;
bp->data = NULL;
break;
}
ASSERT(bp->lmfs_needsetcache == 0);
/* Block needed has been found. */
if (bp->lmfs_count == 0) {
rm_lru(bp);
ASSERT(!(bp->lmfs_flags & VMMC_BLOCK_LOCKED));
bp->lmfs_flags |= VMMC_BLOCK_LOCKED;
}
raisecount(bp);
ASSERT(bp->lmfs_bytes == fs_block_size);
ASSERT(bp->lmfs_dev == dev);
ASSERT(bp->lmfs_dev != NO_DEV);
ASSERT(bp->lmfs_flags & VMMC_BLOCK_LOCKED);
ASSERT(bp->data);
if(ino != VMC_NO_INODE) {
if(bp->lmfs_inode == VMC_NO_INODE
|| bp->lmfs_inode != ino
|| bp->lmfs_inode_offset != ino_off) {
bp->lmfs_inode = ino;
bp->lmfs_inode_offset = ino_off;
bp->lmfs_needsetcache = 1;
}
}
return(bp);
} else {
/* This block is not the one sought. */
bp = bp->lmfs_hash; /* move to next block on hash chain */
}
}
/* Desired block is not on available chain. Find a free block to use. */
if(bp) {
ASSERT(bp->lmfs_flags & VMMC_EVICTED);
} else {
if ((bp = front) == NULL) panic("all buffers in use: %d", nr_bufs);
}
assert(bp);
rm_lru(bp);
/* Remove the block that was just taken from its hash chain. */
b = BUFHASH(bp->lmfs_blocknr);
prev_ptr = buf_hash[b];
if (prev_ptr == bp) {
buf_hash[b] = bp->lmfs_hash;
} else {
/* The block just taken is not on the front of its hash chain. */
while (prev_ptr->lmfs_hash != NULL)
if (prev_ptr->lmfs_hash == bp) {
prev_ptr->lmfs_hash = bp->lmfs_hash; /* found it */
break;
} else {
prev_ptr = prev_ptr->lmfs_hash; /* keep looking */
}
}
freeblock(bp);
bp->lmfs_inode = ino;
bp->lmfs_inode_offset = ino_off;
bp->lmfs_flags = VMMC_BLOCK_LOCKED;
bp->lmfs_needsetcache = 0;
bp->lmfs_dev = dev; /* fill in device number */
bp->lmfs_blocknr = block; /* fill in block number */
ASSERT(bp->lmfs_count == 0);
raisecount(bp);
b = BUFHASH(bp->lmfs_blocknr);
bp->lmfs_hash = buf_hash[b];
buf_hash[b] = bp; /* add to hash list */
assert(dev != NO_DEV);
/* Block is not found in our cache, but we do want it
* if it's in the vm cache.
*/
assert(!bp->data);
assert(!bp->lmfs_bytes);
if(vmcache) {
if((bp->data = vm_map_cacheblock(dev, dev_off, ino, ino_off,
&bp->lmfs_flags, fs_block_size)) != MAP_FAILED) {
bp->lmfs_bytes = fs_block_size;
ASSERT(!bp->lmfs_needsetcache);
return bp;
}
}
bp->data = NULL;
/* Not in the cache; reserve memory for its contents. */
lmfs_alloc_block(bp);
assert(bp->data);
if(only_search == PREFETCH) {
/* PREFETCH: don't do i/o. */
bp->lmfs_dev = NO_DEV;
} else if (only_search == NORMAL) {
read_block(bp);
} else if(only_search == NO_READ) {
/* This block will be overwritten by new contents. */
} else
panic("unexpected only_search value: %d", only_search);
assert(bp->data);
return(bp); /* return the newly acquired block */
}
/*===========================================================================*
* get_block_ino *
*===========================================================================*/
static int get_block_ino(struct buf **bpp, dev_t dev, block64_t block, int how,
ino_t ino, u64_t ino_off, size_t block_size)
{
/* Check to see if the requested block is in the block cache. The requested
* block is identified by the block number in 'block' on device 'dev', counted
* in the file system block size. The amount of data requested for this block
* is given in 'block_size', which may be less than the file system block size
* iff the requested block is the last (partial) block on a device. Note that
* the given block size does *not* affect the conversion of 'block' to a byte
* offset! Either way, if the block could be obtained, either from the cache
* or by reading from the device, return OK, with a pointer to the buffer
* structure stored in 'bpp'. If not, return a negative error code (and no
* buffer). If necessary, evict some other block and fetch the contents from
* disk (if 'how' is NORMAL). If 'how' is NO_READ, the caller intends to
* overwrite the requested block in its entirety, so it is only necessary to
* see if it is in the cache; if it is not, any free buffer will do. If 'how'
* is PEEK, the function returns the block if it is in the cache or the VM
* cache, and an ENOENT error code otherwise.
* In addition to the LRU chain, there is also a hash chain to link together
* blocks whose block numbers end with the same bit strings, for fast lookup.
*/
int b, r;
static struct buf *bp;
uint64_t dev_off;
struct buf *prev_ptr;
assert(buf_hash);
assert(buf);
assert(nr_bufs > 0);
ASSERT(fs_block_size > 0);
assert(dev != NO_DEV);
assert(block <= UINT64_MAX / fs_block_size);
dev_off = block * fs_block_size;
if((ino_off % fs_block_size)) {
printf("cache: unaligned lmfs_get_block_ino ino_off %llu\n",
ino_off);
util_stacktrace();
}
/* See if the block is in the cache. If so, we can return it right away. */
bp = find_block(dev, block);
if (bp != NULL && !(bp->lmfs_flags & VMMC_EVICTED)) {
ASSERT(bp->lmfs_dev == dev);
ASSERT(bp->lmfs_dev != NO_DEV);
/* The block must have exactly the requested number of bytes. */
if (bp->lmfs_bytes != block_size)
return EIO;
/* Block needed has been found. */
if (bp->lmfs_count == 0) {
rm_lru(bp);
ASSERT(bp->lmfs_needsetcache == 0);
ASSERT(!(bp->lmfs_flags & VMMC_BLOCK_LOCKED));
/* FIXME: race condition against the VMMC_EVICTED check */
bp->lmfs_flags |= VMMC_BLOCK_LOCKED;
}
raisecount(bp);
ASSERT(bp->lmfs_flags & VMMC_BLOCK_LOCKED);
ASSERT(bp->data);
if(ino != VMC_NO_INODE) {
if(bp->lmfs_inode == VMC_NO_INODE
|| bp->lmfs_inode != ino
|| bp->lmfs_inode_offset != ino_off) {
bp->lmfs_inode = ino;
bp->lmfs_inode_offset = ino_off;
bp->lmfs_needsetcache = 1;
}
}
*bpp = bp;
return OK;
}
/* We had the block in the cache but VM evicted it; invalidate it. */
if (bp != NULL) {
assert(bp->lmfs_flags & VMMC_EVICTED);
ASSERT(bp->lmfs_count == 0);
ASSERT(!(bp->lmfs_flags & VMMC_BLOCK_LOCKED));
ASSERT(!(bp->lmfs_flags & VMMC_DIRTY));
bp->lmfs_dev = NO_DEV;
bp->lmfs_bytes = 0;
bp->data = NULL;
}
/* Desired block is not on available chain. Find a free block to use. */
if(bp) {
ASSERT(bp->lmfs_flags & VMMC_EVICTED);
} else {
if ((bp = front) == NULL) panic("all buffers in use: %d", nr_bufs);
}
assert(bp);
rm_lru(bp);
/* Remove the block that was just taken from its hash chain. */
b = BUFHASH(bp->lmfs_blocknr);
prev_ptr = buf_hash[b];
if (prev_ptr == bp) {
buf_hash[b] = bp->lmfs_hash;
} else {
/* The block just taken is not on the front of its hash chain. */
while (prev_ptr->lmfs_hash != NULL)
if (prev_ptr->lmfs_hash == bp) {
prev_ptr->lmfs_hash = bp->lmfs_hash; /* found it */
break;
} else {
prev_ptr = prev_ptr->lmfs_hash; /* keep looking */
}
}
freeblock(bp);
bp->lmfs_inode = ino;
bp->lmfs_inode_offset = ino_off;
bp->lmfs_flags = VMMC_BLOCK_LOCKED;
bp->lmfs_needsetcache = 0;
bp->lmfs_dev = dev; /* fill in device number */
bp->lmfs_blocknr = block; /* fill in block number */
ASSERT(bp->lmfs_count == 0);
raisecount(bp);
b = BUFHASH(bp->lmfs_blocknr);
bp->lmfs_hash = buf_hash[b];
buf_hash[b] = bp; /* add to hash list */
assert(dev != NO_DEV);
/* The block is not found in our cache, but we do want it if it's in the VM
* cache. The exception is NO_READ, purely for context switching performance
* reasons. NO_READ is used for 1) newly allocated blocks, 2) blocks being
* prefetched, and 3) blocks about to be fully overwritten. In the first two
* cases, VM will not have the block in its cache anyway, and for the third
* we save on one VM call only if the block is in the VM cache.
*/
assert(!bp->data);
assert(!bp->lmfs_bytes);
if (how != NO_READ && vmcache) {
if((bp->data = vm_map_cacheblock(dev, dev_off, ino, ino_off,
&bp->lmfs_flags, roundup(block_size, PAGE_SIZE))) != MAP_FAILED) {
bp->lmfs_bytes = block_size;
ASSERT(!bp->lmfs_needsetcache);
*bpp = bp;
return OK;
}
}
bp->data = NULL;
/* The block is not in the cache, and VM does not know about it. If we were
* requested to search for the block only, we can now return failure to the
* caller. Return the block to the pool without allocating data pages, since
* these would be freed upon recycling the block anyway.
*/
if (how == PEEK) {
bp->lmfs_dev = NO_DEV;
put_block(bp, ONE_SHOT);
return ENOENT;
}
/* Not in the cache; reserve memory for its contents. */
lmfs_alloc_block(bp, block_size);
assert(bp->data);
if (how == NORMAL) {
/* Try to read the block. Return an error code on failure. */
if ((r = read_block(bp, block_size)) != OK) {
put_block(bp, 0);
return r;
}
} else if(how == NO_READ) {
/* This block will be overwritten by new contents. */
} else
panic("unexpected 'how' value: %d", how);
assert(bp->data);
*bpp = bp; /* return the newly acquired block */
return OK;
}
/*===========================================================================*
* vm_allocpage *
*===========================================================================*/
void *vm_allocpages(phys_bytes *phys, int reason, int pages)
{
/* Allocate a page for use by VM itself. */
phys_bytes newpage;
static int level = 0;
void *ret;
u32_t mem_flags = 0;
assert(reason >= 0 && reason < VMP_CATEGORIES);
assert(pages > 0);
level++;
assert(level >= 1);
assert(level <= 2);
if((level > 1) || !pt_init_done) {
void *s;
if(pages == 1) s=vm_getsparepage(phys);
else if(pages == 4) s=vm_getsparepagedir(phys);
else panic("%d pages", pages);
level--;
if(!s) {
util_stacktrace();
printf("VM: warning: out of spare pages\n");
}
if(!is_staticaddr(s)) vm_self_pages++;
return s;
}
#if defined(__arm__)
if (reason == VMP_PAGEDIR) {
mem_flags |= PAF_ALIGN16K;
}
#endif
/* Allocate page of memory for use by VM. As VM
* is trusted, we don't have to pre-clear it.
*/
if((newpage = alloc_mem(pages, mem_flags)) == NO_MEM) {
level--;
printf("VM: vm_allocpage: alloc_mem failed\n");
return NULL;
}
*phys = CLICK2ABS(newpage);
if(!(ret = vm_mappages(*phys, pages))) {
level--;
printf("VM: vm_allocpage: vm_mappages failed\n");
return NULL;
}
level--;
vm_self_pages++;
return ret;
}
/*===========================================================================*
* vm_allocpage *
*===========================================================================*/
PUBLIC void *vm_allocpage(phys_bytes *phys, int reason)
{
/* Allocate a page for use by VM itself. */
phys_bytes newpage;
vir_bytes loc;
pt_t *pt;
int r;
static int level = 0;
void *ret;
pt = &vmprocess->vm_pt;
assert(reason >= 0 && reason < VMP_CATEGORIES);
level++;
assert(level >= 1);
assert(level <= 2);
if(level > 1 || !(vmprocess->vm_flags & VMF_HASPT) || !meminit_done) {
int r;
void *s;
s=vm_getsparepage(phys);
level--;
if(!s) {
util_stacktrace();
printf("VM: warning: out of spare pages\n");
}
return s;
}
/* VM does have a pagetable, so get a page and map it in there.
* Where in our virtual address space can we put it?
*/
loc = findhole(pt, arch_vir2map(vmprocess, vmprocess->vm_stacktop),
vmprocess->vm_arch.vm_data_top);
if(loc == NO_MEM) {
level--;
printf("VM: vm_allocpage: findhole failed\n");
return NULL;
}
/* Allocate page of memory for use by VM. As VM
* is trusted, we don't have to pre-clear it.
*/
if((newpage = alloc_mem(CLICKSPERPAGE, 0)) == NO_MEM) {
level--;
printf("VM: vm_allocpage: alloc_mem failed\n");
return NULL;
}
*phys = CLICK2ABS(newpage);
/* Map this page into our address space. */
if((r=pt_writemap(vmprocess, pt, loc, *phys, I386_PAGE_SIZE,
I386_VM_PRESENT | I386_VM_USER | I386_VM_WRITE, 0)) != OK) {
free_mem(newpage, CLICKSPERPAGE);
printf("vm_allocpage writemap failed\n");
level--;
return NULL;
}
if((r=sys_vmctl(SELF, VMCTL_FLUSHTLB, 0)) != OK) {
panic("VMCTL_FLUSHTLB failed: %d", r);
}
level--;
/* Return user-space-ready pointer to it. */
ret = (void *) arch_map2vir(vmprocess, loc);
return ret;
}
