linux/lib/idr.c

1165 lines
29 KiB
C
Raw Normal View History

/*
* 2002-10-18 written by Jim Houston jim.houston@ccur.com
* Copyright (C) 2002 by Concurrent Computer Corporation
* Distributed under the GNU GPL license version 2.
*
* Modified by George Anzinger to reuse immediately and to use
* find bit instructions. Also removed _irq on spinlocks.
*
* Modified by Nadia Derbey to make it RCU safe.
*
* Small id to pointer translation service.
*
* It uses a radix tree like structure as a sparse array indexed
* by the id to obtain the pointer. The bitmap makes allocating
* a new id quick.
*
* You call it to allocate an id (an int) an associate with that id a
* pointer or what ever, we treat it as a (void *). You can pass this
* id to a user for him to pass back at a later time. You then pass
* that id to this code and it returns your pointer.
* You can release ids at any time. When all ids are released, most of
* the memory is returned (we keep MAX_IDR_FREE) in a local pool so we
* don't need to go to the memory "store" during an id allocate, just
* so you don't need to be too concerned about locking and conflicts
* with the slab allocator.
*/
#ifndef TEST // to test in user space...
#include <linux/slab.h>
#include <linux/init.h>
#include <linux/export.h>
#endif
#include <linux/err.h>
#include <linux/string.h>
#include <linux/idr.h>
#include <linux/spinlock.h>
idr: implement idr_preload[_end]() and idr_alloc() The current idr interface is very cumbersome. * For all allocations, two function calls - idr_pre_get() and idr_get_new*() - should be made. * idr_pre_get() doesn't guarantee that the following idr_get_new*() will not fail from memory shortage. If idr_get_new*() returns -EAGAIN, the caller is expected to retry pre_get and allocation. * idr_get_new*() can't enforce upper limit. Upper limit can only be enforced by allocating and then freeing if above limit. * idr_layer buffer is unnecessarily per-idr. Each idr ends up keeping around MAX_IDR_FREE idr_layers. The memory consumed per idr is under two pages but it makes it difficult to make idr_layer larger. This patch implements the following new set of allocation functions. * idr_preload[_end]() - Similar to radix preload but doesn't fail. The first idr_alloc() inside preload section can be treated as if it were called with @gfp_mask used for idr_preload(). * idr_alloc() - Allocate an ID w/ lower and upper limits. Takes @gfp_flags and can be used w/o preloading. When used inside preloaded section, the allocation mask of preloading can be assumed. If idr_alloc() can be called from a context which allows sufficiently relaxed @gfp_mask, it can be used by itself. If, for example, idr_alloc() is called inside spinlock protected region, preloading can be used like the following. idr_preload(GFP_KERNEL); spin_lock(lock); id = idr_alloc(idr, ptr, start, end, GFP_NOWAIT); spin_unlock(lock); idr_preload_end(); if (id < 0) error; which is much simpler and less error-prone than idr_pre_get and idr_get_new*() loop. The new interface uses per-pcu idr_layer buffer and thus the number of idr's in the system doesn't affect the amount of memory used for preloading. idr_layer_alloc() is introduced to handle idr_layer allocations for both old and new ID allocation paths. This is a bit hairy now but the new interface is expected to replace the old and the internal implementation eventually will become simpler. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:03:55 +00:00
#include <linux/percpu.h>
#include <linux/hardirq.h>
idr: remove MAX_IDR_MASK and move left MAX_IDR_* into idr.c MAX_IDR_MASK is another weirdness in the idr interface. As idr covers whole positive integer range, it's defined as 0x7fffffff or INT_MAX. Its usage in idr_find(), idr_replace() and idr_remove() is bizarre. They basically mask off the sign bit and operate on the rest, so if the caller, by accident, passes in a negative number, the sign bit will be masked off and the remaining part will be used as if that was the input, which is worse than crashing. The constant is visible in idr.h and there are several users in the kernel. * drivers/i2c/i2c-core.c:i2c_add_numbered_adapter() Basically used to test if adap->nr is a negative number which isn't -1 and returns -EINVAL if so. idr_alloc() already has negative @start checking (w/ WARN_ON_ONCE), so this can go away. * drivers/infiniband/core/cm.c:cm_alloc_id() drivers/infiniband/hw/mlx4/cm.c:id_map_alloc() Used to wrap cyclic @start. Can be replaced with max(next, 0). Note that this type of cyclic allocation using idr is buggy. These are prone to spurious -ENOSPC failure after the first wraparound. * fs/super.c:get_anon_bdev() The ID allocated from ida is masked off before being tested whether it's inside valid range. ida allocated ID can never be a negative number and the masking is unnecessary. Update idr_*() functions to fail with -EINVAL when negative @id is specified and update other MAX_IDR_MASK users as described above. This leaves MAX_IDR_MASK without any user, remove it and relocate other MAX_IDR_* constants to lib/idr.c. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Roland Dreier <roland@kernel.org> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: "Marciniszyn, Mike" <mike.marciniszyn@intel.com> Cc: Jack Morgenstein <jackm@dev.mellanox.co.il> Cc: Or Gerlitz <ogerlitz@mellanox.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Acked-by: Wolfram Sang <wolfram@the-dreams.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:05:04 +00:00
#define MAX_IDR_SHIFT (sizeof(int) * 8 - 1)
#define MAX_IDR_BIT (1U << MAX_IDR_SHIFT)
/* Leave the possibility of an incomplete final layer */
#define MAX_IDR_LEVEL ((MAX_IDR_SHIFT + IDR_BITS - 1) / IDR_BITS)
/* Number of id_layer structs to leave in free list */
#define MAX_IDR_FREE (MAX_IDR_LEVEL * 2)
static struct kmem_cache *idr_layer_cache;
idr: implement idr_preload[_end]() and idr_alloc() The current idr interface is very cumbersome. * For all allocations, two function calls - idr_pre_get() and idr_get_new*() - should be made. * idr_pre_get() doesn't guarantee that the following idr_get_new*() will not fail from memory shortage. If idr_get_new*() returns -EAGAIN, the caller is expected to retry pre_get and allocation. * idr_get_new*() can't enforce upper limit. Upper limit can only be enforced by allocating and then freeing if above limit. * idr_layer buffer is unnecessarily per-idr. Each idr ends up keeping around MAX_IDR_FREE idr_layers. The memory consumed per idr is under two pages but it makes it difficult to make idr_layer larger. This patch implements the following new set of allocation functions. * idr_preload[_end]() - Similar to radix preload but doesn't fail. The first idr_alloc() inside preload section can be treated as if it were called with @gfp_mask used for idr_preload(). * idr_alloc() - Allocate an ID w/ lower and upper limits. Takes @gfp_flags and can be used w/o preloading. When used inside preloaded section, the allocation mask of preloading can be assumed. If idr_alloc() can be called from a context which allows sufficiently relaxed @gfp_mask, it can be used by itself. If, for example, idr_alloc() is called inside spinlock protected region, preloading can be used like the following. idr_preload(GFP_KERNEL); spin_lock(lock); id = idr_alloc(idr, ptr, start, end, GFP_NOWAIT); spin_unlock(lock); idr_preload_end(); if (id < 0) error; which is much simpler and less error-prone than idr_pre_get and idr_get_new*() loop. The new interface uses per-pcu idr_layer buffer and thus the number of idr's in the system doesn't affect the amount of memory used for preloading. idr_layer_alloc() is introduced to handle idr_layer allocations for both old and new ID allocation paths. This is a bit hairy now but the new interface is expected to replace the old and the internal implementation eventually will become simpler. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:03:55 +00:00
static DEFINE_PER_CPU(struct idr_layer *, idr_preload_head);
static DEFINE_PER_CPU(int, idr_preload_cnt);
static DEFINE_SPINLOCK(simple_ida_lock);
idr: fix top layer handling Most functions in idr fail to deal with the high bits when the idr tree grows to the maximum height. * idr_get_empty_slot() stops growing idr tree once the depth reaches MAX_IDR_LEVEL - 1, which is one depth shallower than necessary to cover the whole range. The function doesn't even notice that it didn't grow the tree enough and ends up allocating the wrong ID given sufficiently high @starting_id. For example, on 64 bit, if the starting id is 0x7fffff01, idr_get_empty_slot() will grow the tree 5 layer deep, which only covers the 30 bits and then proceed to allocate as if the bit 30 wasn't specified. It ends up allocating 0x3fffff01 without the bit 30 but still returns 0x7fffff01. * __idr_remove_all() will not remove anything if the tree is fully grown. * idr_find() can't find anything if the tree is fully grown. * idr_for_each() and idr_get_next() can't iterate anything if the tree is fully grown. Fix it by introducing idr_max() which returns the maximum possible ID given the depth of tree and replacing the id limit checks in all affected places. As the idr_layer pointer array pa[] needs to be 1 larger than the maximum depth, enlarge pa[] arrays by one. While this plugs the discovered issues, the whole code base is horrible and in desparate need of rewrite. It's fragile like hell, Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:05:02 +00:00
/* the maximum ID which can be allocated given idr->layers */
static int idr_max(int layers)
{
int bits = min_t(int, layers * IDR_BITS, MAX_IDR_SHIFT);
return (1 << bits) - 1;
}
/*
* Prefix mask for an idr_layer at @layer. For layer 0, the prefix mask is
* all bits except for the lower IDR_BITS. For layer 1, 2 * IDR_BITS, and
* so on.
*/
static int idr_layer_prefix_mask(int layer)
{
return ~idr_max(layer + 1);
}
static struct idr_layer *get_from_free_list(struct idr *idp)
{
struct idr_layer *p;
unsigned long flags;
spin_lock_irqsave(&idp->lock, flags);
if ((p = idp->id_free)) {
idp->id_free = p->ary[0];
idp->id_free_cnt--;
p->ary[0] = NULL;
}
spin_unlock_irqrestore(&idp->lock, flags);
return(p);
}
idr: implement idr_preload[_end]() and idr_alloc() The current idr interface is very cumbersome. * For all allocations, two function calls - idr_pre_get() and idr_get_new*() - should be made. * idr_pre_get() doesn't guarantee that the following idr_get_new*() will not fail from memory shortage. If idr_get_new*() returns -EAGAIN, the caller is expected to retry pre_get and allocation. * idr_get_new*() can't enforce upper limit. Upper limit can only be enforced by allocating and then freeing if above limit. * idr_layer buffer is unnecessarily per-idr. Each idr ends up keeping around MAX_IDR_FREE idr_layers. The memory consumed per idr is under two pages but it makes it difficult to make idr_layer larger. This patch implements the following new set of allocation functions. * idr_preload[_end]() - Similar to radix preload but doesn't fail. The first idr_alloc() inside preload section can be treated as if it were called with @gfp_mask used for idr_preload(). * idr_alloc() - Allocate an ID w/ lower and upper limits. Takes @gfp_flags and can be used w/o preloading. When used inside preloaded section, the allocation mask of preloading can be assumed. If idr_alloc() can be called from a context which allows sufficiently relaxed @gfp_mask, it can be used by itself. If, for example, idr_alloc() is called inside spinlock protected region, preloading can be used like the following. idr_preload(GFP_KERNEL); spin_lock(lock); id = idr_alloc(idr, ptr, start, end, GFP_NOWAIT); spin_unlock(lock); idr_preload_end(); if (id < 0) error; which is much simpler and less error-prone than idr_pre_get and idr_get_new*() loop. The new interface uses per-pcu idr_layer buffer and thus the number of idr's in the system doesn't affect the amount of memory used for preloading. idr_layer_alloc() is introduced to handle idr_layer allocations for both old and new ID allocation paths. This is a bit hairy now but the new interface is expected to replace the old and the internal implementation eventually will become simpler. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:03:55 +00:00
/**
* idr_layer_alloc - allocate a new idr_layer
* @gfp_mask: allocation mask
* @layer_idr: optional idr to allocate from
*
* If @layer_idr is %NULL, directly allocate one using @gfp_mask or fetch
* one from the per-cpu preload buffer. If @layer_idr is not %NULL, fetch
* an idr_layer from @idr->id_free.
*
* @layer_idr is to maintain backward compatibility with the old alloc
* interface - idr_pre_get() and idr_get_new*() - and will be removed
* together with per-pool preload buffer.
*/
static struct idr_layer *idr_layer_alloc(gfp_t gfp_mask, struct idr *layer_idr)
{
struct idr_layer *new;
/* this is the old path, bypass to get_from_free_list() */
if (layer_idr)
return get_from_free_list(layer_idr);
/*
* Try to allocate directly from kmem_cache. We want to try this
* before preload buffer; otherwise, non-preloading idr_alloc()
* users will end up taking advantage of preloading ones. As the
* following is allowed to fail for preloaded cases, suppress
* warning this time.
*/
new = kmem_cache_zalloc(idr_layer_cache, gfp_mask | __GFP_NOWARN);
idr: implement idr_preload[_end]() and idr_alloc() The current idr interface is very cumbersome. * For all allocations, two function calls - idr_pre_get() and idr_get_new*() - should be made. * idr_pre_get() doesn't guarantee that the following idr_get_new*() will not fail from memory shortage. If idr_get_new*() returns -EAGAIN, the caller is expected to retry pre_get and allocation. * idr_get_new*() can't enforce upper limit. Upper limit can only be enforced by allocating and then freeing if above limit. * idr_layer buffer is unnecessarily per-idr. Each idr ends up keeping around MAX_IDR_FREE idr_layers. The memory consumed per idr is under two pages but it makes it difficult to make idr_layer larger. This patch implements the following new set of allocation functions. * idr_preload[_end]() - Similar to radix preload but doesn't fail. The first idr_alloc() inside preload section can be treated as if it were called with @gfp_mask used for idr_preload(). * idr_alloc() - Allocate an ID w/ lower and upper limits. Takes @gfp_flags and can be used w/o preloading. When used inside preloaded section, the allocation mask of preloading can be assumed. If idr_alloc() can be called from a context which allows sufficiently relaxed @gfp_mask, it can be used by itself. If, for example, idr_alloc() is called inside spinlock protected region, preloading can be used like the following. idr_preload(GFP_KERNEL); spin_lock(lock); id = idr_alloc(idr, ptr, start, end, GFP_NOWAIT); spin_unlock(lock); idr_preload_end(); if (id < 0) error; which is much simpler and less error-prone than idr_pre_get and idr_get_new*() loop. The new interface uses per-pcu idr_layer buffer and thus the number of idr's in the system doesn't affect the amount of memory used for preloading. idr_layer_alloc() is introduced to handle idr_layer allocations for both old and new ID allocation paths. This is a bit hairy now but the new interface is expected to replace the old and the internal implementation eventually will become simpler. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:03:55 +00:00
if (new)
return new;
/*
* Try to fetch one from the per-cpu preload buffer if in process
* context. See idr_preload() for details.
*/
if (!in_interrupt()) {
preempt_disable();
new = __this_cpu_read(idr_preload_head);
if (new) {
__this_cpu_write(idr_preload_head, new->ary[0]);
__this_cpu_dec(idr_preload_cnt);
new->ary[0] = NULL;
}
preempt_enable();
if (new)
return new;
idr: implement idr_preload[_end]() and idr_alloc() The current idr interface is very cumbersome. * For all allocations, two function calls - idr_pre_get() and idr_get_new*() - should be made. * idr_pre_get() doesn't guarantee that the following idr_get_new*() will not fail from memory shortage. If idr_get_new*() returns -EAGAIN, the caller is expected to retry pre_get and allocation. * idr_get_new*() can't enforce upper limit. Upper limit can only be enforced by allocating and then freeing if above limit. * idr_layer buffer is unnecessarily per-idr. Each idr ends up keeping around MAX_IDR_FREE idr_layers. The memory consumed per idr is under two pages but it makes it difficult to make idr_layer larger. This patch implements the following new set of allocation functions. * idr_preload[_end]() - Similar to radix preload but doesn't fail. The first idr_alloc() inside preload section can be treated as if it were called with @gfp_mask used for idr_preload(). * idr_alloc() - Allocate an ID w/ lower and upper limits. Takes @gfp_flags and can be used w/o preloading. When used inside preloaded section, the allocation mask of preloading can be assumed. If idr_alloc() can be called from a context which allows sufficiently relaxed @gfp_mask, it can be used by itself. If, for example, idr_alloc() is called inside spinlock protected region, preloading can be used like the following. idr_preload(GFP_KERNEL); spin_lock(lock); id = idr_alloc(idr, ptr, start, end, GFP_NOWAIT); spin_unlock(lock); idr_preload_end(); if (id < 0) error; which is much simpler and less error-prone than idr_pre_get and idr_get_new*() loop. The new interface uses per-pcu idr_layer buffer and thus the number of idr's in the system doesn't affect the amount of memory used for preloading. idr_layer_alloc() is introduced to handle idr_layer allocations for both old and new ID allocation paths. This is a bit hairy now but the new interface is expected to replace the old and the internal implementation eventually will become simpler. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:03:55 +00:00
}
/*
* Both failed. Try kmem_cache again w/o adding __GFP_NOWARN so
* that memory allocation failure warning is printed as intended.
*/
return kmem_cache_zalloc(idr_layer_cache, gfp_mask);
idr: implement idr_preload[_end]() and idr_alloc() The current idr interface is very cumbersome. * For all allocations, two function calls - idr_pre_get() and idr_get_new*() - should be made. * idr_pre_get() doesn't guarantee that the following idr_get_new*() will not fail from memory shortage. If idr_get_new*() returns -EAGAIN, the caller is expected to retry pre_get and allocation. * idr_get_new*() can't enforce upper limit. Upper limit can only be enforced by allocating and then freeing if above limit. * idr_layer buffer is unnecessarily per-idr. Each idr ends up keeping around MAX_IDR_FREE idr_layers. The memory consumed per idr is under two pages but it makes it difficult to make idr_layer larger. This patch implements the following new set of allocation functions. * idr_preload[_end]() - Similar to radix preload but doesn't fail. The first idr_alloc() inside preload section can be treated as if it were called with @gfp_mask used for idr_preload(). * idr_alloc() - Allocate an ID w/ lower and upper limits. Takes @gfp_flags and can be used w/o preloading. When used inside preloaded section, the allocation mask of preloading can be assumed. If idr_alloc() can be called from a context which allows sufficiently relaxed @gfp_mask, it can be used by itself. If, for example, idr_alloc() is called inside spinlock protected region, preloading can be used like the following. idr_preload(GFP_KERNEL); spin_lock(lock); id = idr_alloc(idr, ptr, start, end, GFP_NOWAIT); spin_unlock(lock); idr_preload_end(); if (id < 0) error; which is much simpler and less error-prone than idr_pre_get and idr_get_new*() loop. The new interface uses per-pcu idr_layer buffer and thus the number of idr's in the system doesn't affect the amount of memory used for preloading. idr_layer_alloc() is introduced to handle idr_layer allocations for both old and new ID allocation paths. This is a bit hairy now but the new interface is expected to replace the old and the internal implementation eventually will become simpler. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:03:55 +00:00
}
static void idr_layer_rcu_free(struct rcu_head *head)
{
struct idr_layer *layer;
layer = container_of(head, struct idr_layer, rcu_head);
kmem_cache_free(idr_layer_cache, layer);
}
static inline void free_layer(struct idr *idr, struct idr_layer *p)
{
if (idr->hint && idr->hint == p)
RCU_INIT_POINTER(idr->hint, NULL);
call_rcu(&p->rcu_head, idr_layer_rcu_free);
}
/* only called when idp->lock is held */
static void __move_to_free_list(struct idr *idp, struct idr_layer *p)
{
p->ary[0] = idp->id_free;
idp->id_free = p;
idp->id_free_cnt++;
}
static void move_to_free_list(struct idr *idp, struct idr_layer *p)
{
unsigned long flags;
/*
* Depends on the return element being zeroed.
*/
spin_lock_irqsave(&idp->lock, flags);
__move_to_free_list(idp, p);
spin_unlock_irqrestore(&idp->lock, flags);
}
static void idr_mark_full(struct idr_layer **pa, int id)
{
struct idr_layer *p = pa[0];
int l = 0;
__set_bit(id & IDR_MASK, p->bitmap);
/*
* If this layer is full mark the bit in the layer above to
* show that this part of the radix tree is full. This may
* complete the layer above and require walking up the radix
* tree.
*/
while (bitmap_full(p->bitmap, IDR_SIZE)) {
if (!(p = pa[++l]))
break;
id = id >> IDR_BITS;
__set_bit((id & IDR_MASK), p->bitmap);
}
}
int __idr_pre_get(struct idr *idp, gfp_t gfp_mask)
{
while (idp->id_free_cnt < MAX_IDR_FREE) {
struct idr_layer *new;
lib/idr.c: use kmem_cache_zalloc() for the idr_layer cache David points out that the idr_remove_all() function returns unused slabs to the kmem cache, but needs to zero them first or else they will be uninitialized upon next use. This causes crashes which have been observed in the firewire subsystem. He fixed this by zeroing the object before freeing it in idr_remove_all(). But we agree that simply removing the constructor and zeroing the object at allocation time is simpler than relying upon slab constructor machinery and might even be faster. This problem was introduced by "idr: make idr_remove rcu-safe" (commit cf481c20c476ad2c0febdace9ce23f5a4db19582), which was first released in 2.6.27. There are no known codesites which trigger this bug in 2.6.27 or 2.6.28. The post-2.6.28 firewire changes are the only known triggerer. There might of course be not-yet-discovered triggerers in 2.6.27 and 2.6.28, and there might be out-of-tree triggerers which are added to those kernel versions. I'll let the -stable guys decide whether they want to backport this fix. Reported-by: David Moore <dcm@acm.org> Cc: Stefan Richter <stefanr@s5r6.in-berlin.de> Cc: Nadia Derbey <Nadia.Derbey@bull.net> Cc: Paul E. McKenney <paulmck@us.ibm.com> Cc: Manfred Spraul <manfred@colorfullife.com> Cc: Kristian Hgsberg <krh@redhat.com> Acked-by: Pekka Enberg <penberg@cs.helsinki.fi> Cc: <stable@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-01-15 21:51:21 +00:00
new = kmem_cache_zalloc(idr_layer_cache, gfp_mask);
if (new == NULL)
return (0);
move_to_free_list(idp, new);
}
return 1;
}
EXPORT_SYMBOL(__idr_pre_get);
/**
* sub_alloc - try to allocate an id without growing the tree depth
* @idp: idr handle
* @starting_id: id to start search at
* @pa: idr_layer[MAX_IDR_LEVEL] used as backtrack buffer
idr: implement idr_preload[_end]() and idr_alloc() The current idr interface is very cumbersome. * For all allocations, two function calls - idr_pre_get() and idr_get_new*() - should be made. * idr_pre_get() doesn't guarantee that the following idr_get_new*() will not fail from memory shortage. If idr_get_new*() returns -EAGAIN, the caller is expected to retry pre_get and allocation. * idr_get_new*() can't enforce upper limit. Upper limit can only be enforced by allocating and then freeing if above limit. * idr_layer buffer is unnecessarily per-idr. Each idr ends up keeping around MAX_IDR_FREE idr_layers. The memory consumed per idr is under two pages but it makes it difficult to make idr_layer larger. This patch implements the following new set of allocation functions. * idr_preload[_end]() - Similar to radix preload but doesn't fail. The first idr_alloc() inside preload section can be treated as if it were called with @gfp_mask used for idr_preload(). * idr_alloc() - Allocate an ID w/ lower and upper limits. Takes @gfp_flags and can be used w/o preloading. When used inside preloaded section, the allocation mask of preloading can be assumed. If idr_alloc() can be called from a context which allows sufficiently relaxed @gfp_mask, it can be used by itself. If, for example, idr_alloc() is called inside spinlock protected region, preloading can be used like the following. idr_preload(GFP_KERNEL); spin_lock(lock); id = idr_alloc(idr, ptr, start, end, GFP_NOWAIT); spin_unlock(lock); idr_preload_end(); if (id < 0) error; which is much simpler and less error-prone than idr_pre_get and idr_get_new*() loop. The new interface uses per-pcu idr_layer buffer and thus the number of idr's in the system doesn't affect the amount of memory used for preloading. idr_layer_alloc() is introduced to handle idr_layer allocations for both old and new ID allocation paths. This is a bit hairy now but the new interface is expected to replace the old and the internal implementation eventually will become simpler. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:03:55 +00:00
* @gfp_mask: allocation mask for idr_layer_alloc()
* @layer_idr: optional idr passed to idr_layer_alloc()
*
* Allocate an id in range [@starting_id, INT_MAX] from @idp without
* growing its depth. Returns
*
* the allocated id >= 0 if successful,
* -EAGAIN if the tree needs to grow for allocation to succeed,
* -ENOSPC if the id space is exhausted,
* -ENOMEM if more idr_layers need to be allocated.
*/
idr: implement idr_preload[_end]() and idr_alloc() The current idr interface is very cumbersome. * For all allocations, two function calls - idr_pre_get() and idr_get_new*() - should be made. * idr_pre_get() doesn't guarantee that the following idr_get_new*() will not fail from memory shortage. If idr_get_new*() returns -EAGAIN, the caller is expected to retry pre_get and allocation. * idr_get_new*() can't enforce upper limit. Upper limit can only be enforced by allocating and then freeing if above limit. * idr_layer buffer is unnecessarily per-idr. Each idr ends up keeping around MAX_IDR_FREE idr_layers. The memory consumed per idr is under two pages but it makes it difficult to make idr_layer larger. This patch implements the following new set of allocation functions. * idr_preload[_end]() - Similar to radix preload but doesn't fail. The first idr_alloc() inside preload section can be treated as if it were called with @gfp_mask used for idr_preload(). * idr_alloc() - Allocate an ID w/ lower and upper limits. Takes @gfp_flags and can be used w/o preloading. When used inside preloaded section, the allocation mask of preloading can be assumed. If idr_alloc() can be called from a context which allows sufficiently relaxed @gfp_mask, it can be used by itself. If, for example, idr_alloc() is called inside spinlock protected region, preloading can be used like the following. idr_preload(GFP_KERNEL); spin_lock(lock); id = idr_alloc(idr, ptr, start, end, GFP_NOWAIT); spin_unlock(lock); idr_preload_end(); if (id < 0) error; which is much simpler and less error-prone than idr_pre_get and idr_get_new*() loop. The new interface uses per-pcu idr_layer buffer and thus the number of idr's in the system doesn't affect the amount of memory used for preloading. idr_layer_alloc() is introduced to handle idr_layer allocations for both old and new ID allocation paths. This is a bit hairy now but the new interface is expected to replace the old and the internal implementation eventually will become simpler. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:03:55 +00:00
static int sub_alloc(struct idr *idp, int *starting_id, struct idr_layer **pa,
gfp_t gfp_mask, struct idr *layer_idr)
{
int n, m, sh;
struct idr_layer *p, *new;
int l, id, oid;
id = *starting_id;
restart:
p = idp->top;
l = idp->layers;
pa[l--] = NULL;
while (1) {
/*
* We run around this while until we reach the leaf node...
*/
n = (id >> (IDR_BITS*l)) & IDR_MASK;
m = find_next_zero_bit(p->bitmap, IDR_SIZE, n);
if (m == IDR_SIZE) {
/* no space available go back to previous layer. */
l++;
oid = id;
id = (id | ((1 << (IDR_BITS * l)) - 1)) + 1;
/* if already at the top layer, we need to grow */
if (id >= 1 << (idp->layers * IDR_BITS)) {
*starting_id = id;
return -EAGAIN;
}
p = pa[l];
BUG_ON(!p);
/* If we need to go up one layer, continue the
* loop; otherwise, restart from the top.
*/
sh = IDR_BITS * (l + 1);
if (oid >> sh == id >> sh)
continue;
else
goto restart;
}
if (m != n) {
sh = IDR_BITS*l;
id = ((id >> sh) ^ n ^ m) << sh;
}
if ((id >= MAX_IDR_BIT) || (id < 0))
return -ENOSPC;
if (l == 0)
break;
/*
* Create the layer below if it is missing.
*/
if (!p->ary[m]) {
idr: implement idr_preload[_end]() and idr_alloc() The current idr interface is very cumbersome. * For all allocations, two function calls - idr_pre_get() and idr_get_new*() - should be made. * idr_pre_get() doesn't guarantee that the following idr_get_new*() will not fail from memory shortage. If idr_get_new*() returns -EAGAIN, the caller is expected to retry pre_get and allocation. * idr_get_new*() can't enforce upper limit. Upper limit can only be enforced by allocating and then freeing if above limit. * idr_layer buffer is unnecessarily per-idr. Each idr ends up keeping around MAX_IDR_FREE idr_layers. The memory consumed per idr is under two pages but it makes it difficult to make idr_layer larger. This patch implements the following new set of allocation functions. * idr_preload[_end]() - Similar to radix preload but doesn't fail. The first idr_alloc() inside preload section can be treated as if it were called with @gfp_mask used for idr_preload(). * idr_alloc() - Allocate an ID w/ lower and upper limits. Takes @gfp_flags and can be used w/o preloading. When used inside preloaded section, the allocation mask of preloading can be assumed. If idr_alloc() can be called from a context which allows sufficiently relaxed @gfp_mask, it can be used by itself. If, for example, idr_alloc() is called inside spinlock protected region, preloading can be used like the following. idr_preload(GFP_KERNEL); spin_lock(lock); id = idr_alloc(idr, ptr, start, end, GFP_NOWAIT); spin_unlock(lock); idr_preload_end(); if (id < 0) error; which is much simpler and less error-prone than idr_pre_get and idr_get_new*() loop. The new interface uses per-pcu idr_layer buffer and thus the number of idr's in the system doesn't affect the amount of memory used for preloading. idr_layer_alloc() is introduced to handle idr_layer allocations for both old and new ID allocation paths. This is a bit hairy now but the new interface is expected to replace the old and the internal implementation eventually will become simpler. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:03:55 +00:00
new = idr_layer_alloc(gfp_mask, layer_idr);
if (!new)
return -ENOMEM;
new->layer = l-1;
new->prefix = id & idr_layer_prefix_mask(new->layer);
rcu_assign_pointer(p->ary[m], new);
p->count++;
}
pa[l--] = p;
p = p->ary[m];
}
pa[l] = p;
return id;
}
static int idr_get_empty_slot(struct idr *idp, int starting_id,
idr: implement idr_preload[_end]() and idr_alloc() The current idr interface is very cumbersome. * For all allocations, two function calls - idr_pre_get() and idr_get_new*() - should be made. * idr_pre_get() doesn't guarantee that the following idr_get_new*() will not fail from memory shortage. If idr_get_new*() returns -EAGAIN, the caller is expected to retry pre_get and allocation. * idr_get_new*() can't enforce upper limit. Upper limit can only be enforced by allocating and then freeing if above limit. * idr_layer buffer is unnecessarily per-idr. Each idr ends up keeping around MAX_IDR_FREE idr_layers. The memory consumed per idr is under two pages but it makes it difficult to make idr_layer larger. This patch implements the following new set of allocation functions. * idr_preload[_end]() - Similar to radix preload but doesn't fail. The first idr_alloc() inside preload section can be treated as if it were called with @gfp_mask used for idr_preload(). * idr_alloc() - Allocate an ID w/ lower and upper limits. Takes @gfp_flags and can be used w/o preloading. When used inside preloaded section, the allocation mask of preloading can be assumed. If idr_alloc() can be called from a context which allows sufficiently relaxed @gfp_mask, it can be used by itself. If, for example, idr_alloc() is called inside spinlock protected region, preloading can be used like the following. idr_preload(GFP_KERNEL); spin_lock(lock); id = idr_alloc(idr, ptr, start, end, GFP_NOWAIT); spin_unlock(lock); idr_preload_end(); if (id < 0) error; which is much simpler and less error-prone than idr_pre_get and idr_get_new*() loop. The new interface uses per-pcu idr_layer buffer and thus the number of idr's in the system doesn't affect the amount of memory used for preloading. idr_layer_alloc() is introduced to handle idr_layer allocations for both old and new ID allocation paths. This is a bit hairy now but the new interface is expected to replace the old and the internal implementation eventually will become simpler. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:03:55 +00:00
struct idr_layer **pa, gfp_t gfp_mask,
struct idr *layer_idr)
{
struct idr_layer *p, *new;
int layers, v, id;
unsigned long flags;
id = starting_id;
build_up:
p = idp->top;
layers = idp->layers;
if (unlikely(!p)) {
idr: implement idr_preload[_end]() and idr_alloc() The current idr interface is very cumbersome. * For all allocations, two function calls - idr_pre_get() and idr_get_new*() - should be made. * idr_pre_get() doesn't guarantee that the following idr_get_new*() will not fail from memory shortage. If idr_get_new*() returns -EAGAIN, the caller is expected to retry pre_get and allocation. * idr_get_new*() can't enforce upper limit. Upper limit can only be enforced by allocating and then freeing if above limit. * idr_layer buffer is unnecessarily per-idr. Each idr ends up keeping around MAX_IDR_FREE idr_layers. The memory consumed per idr is under two pages but it makes it difficult to make idr_layer larger. This patch implements the following new set of allocation functions. * idr_preload[_end]() - Similar to radix preload but doesn't fail. The first idr_alloc() inside preload section can be treated as if it were called with @gfp_mask used for idr_preload(). * idr_alloc() - Allocate an ID w/ lower and upper limits. Takes @gfp_flags and can be used w/o preloading. When used inside preloaded section, the allocation mask of preloading can be assumed. If idr_alloc() can be called from a context which allows sufficiently relaxed @gfp_mask, it can be used by itself. If, for example, idr_alloc() is called inside spinlock protected region, preloading can be used like the following. idr_preload(GFP_KERNEL); spin_lock(lock); id = idr_alloc(idr, ptr, start, end, GFP_NOWAIT); spin_unlock(lock); idr_preload_end(); if (id < 0) error; which is much simpler and less error-prone than idr_pre_get and idr_get_new*() loop. The new interface uses per-pcu idr_layer buffer and thus the number of idr's in the system doesn't affect the amount of memory used for preloading. idr_layer_alloc() is introduced to handle idr_layer allocations for both old and new ID allocation paths. This is a bit hairy now but the new interface is expected to replace the old and the internal implementation eventually will become simpler. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:03:55 +00:00
if (!(p = idr_layer_alloc(gfp_mask, layer_idr)))
return -ENOMEM;
p->layer = 0;
layers = 1;
}
/*
* Add a new layer to the top of the tree if the requested
* id is larger than the currently allocated space.
*/
idr: fix top layer handling Most functions in idr fail to deal with the high bits when the idr tree grows to the maximum height. * idr_get_empty_slot() stops growing idr tree once the depth reaches MAX_IDR_LEVEL - 1, which is one depth shallower than necessary to cover the whole range. The function doesn't even notice that it didn't grow the tree enough and ends up allocating the wrong ID given sufficiently high @starting_id. For example, on 64 bit, if the starting id is 0x7fffff01, idr_get_empty_slot() will grow the tree 5 layer deep, which only covers the 30 bits and then proceed to allocate as if the bit 30 wasn't specified. It ends up allocating 0x3fffff01 without the bit 30 but still returns 0x7fffff01. * __idr_remove_all() will not remove anything if the tree is fully grown. * idr_find() can't find anything if the tree is fully grown. * idr_for_each() and idr_get_next() can't iterate anything if the tree is fully grown. Fix it by introducing idr_max() which returns the maximum possible ID given the depth of tree and replacing the id limit checks in all affected places. As the idr_layer pointer array pa[] needs to be 1 larger than the maximum depth, enlarge pa[] arrays by one. While this plugs the discovered issues, the whole code base is horrible and in desparate need of rewrite. It's fragile like hell, Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:05:02 +00:00
while (id > idr_max(layers)) {
layers++;
if (!p->count) {
/* special case: if the tree is currently empty,
* then we grow the tree by moving the top node
* upwards.
*/
p->layer++;
WARN_ON_ONCE(p->prefix);
continue;
}
idr: implement idr_preload[_end]() and idr_alloc() The current idr interface is very cumbersome. * For all allocations, two function calls - idr_pre_get() and idr_get_new*() - should be made. * idr_pre_get() doesn't guarantee that the following idr_get_new*() will not fail from memory shortage. If idr_get_new*() returns -EAGAIN, the caller is expected to retry pre_get and allocation. * idr_get_new*() can't enforce upper limit. Upper limit can only be enforced by allocating and then freeing if above limit. * idr_layer buffer is unnecessarily per-idr. Each idr ends up keeping around MAX_IDR_FREE idr_layers. The memory consumed per idr is under two pages but it makes it difficult to make idr_layer larger. This patch implements the following new set of allocation functions. * idr_preload[_end]() - Similar to radix preload but doesn't fail. The first idr_alloc() inside preload section can be treated as if it were called with @gfp_mask used for idr_preload(). * idr_alloc() - Allocate an ID w/ lower and upper limits. Takes @gfp_flags and can be used w/o preloading. When used inside preloaded section, the allocation mask of preloading can be assumed. If idr_alloc() can be called from a context which allows sufficiently relaxed @gfp_mask, it can be used by itself. If, for example, idr_alloc() is called inside spinlock protected region, preloading can be used like the following. idr_preload(GFP_KERNEL); spin_lock(lock); id = idr_alloc(idr, ptr, start, end, GFP_NOWAIT); spin_unlock(lock); idr_preload_end(); if (id < 0) error; which is much simpler and less error-prone than idr_pre_get and idr_get_new*() loop. The new interface uses per-pcu idr_layer buffer and thus the number of idr's in the system doesn't affect the amount of memory used for preloading. idr_layer_alloc() is introduced to handle idr_layer allocations for both old and new ID allocation paths. This is a bit hairy now but the new interface is expected to replace the old and the internal implementation eventually will become simpler. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:03:55 +00:00
if (!(new = idr_layer_alloc(gfp_mask, layer_idr))) {
/*
* The allocation failed. If we built part of
* the structure tear it down.
*/
spin_lock_irqsave(&idp->lock, flags);
for (new = p; p && p != idp->top; new = p) {
p = p->ary[0];
new->ary[0] = NULL;
new->count = 0;
bitmap_clear(new->bitmap, 0, IDR_SIZE);
__move_to_free_list(idp, new);
}
spin_unlock_irqrestore(&idp->lock, flags);
return -ENOMEM;
}
new->ary[0] = p;
new->count = 1;
new->layer = layers-1;
new->prefix = id & idr_layer_prefix_mask(new->layer);
if (bitmap_full(p->bitmap, IDR_SIZE))
__set_bit(0, new->bitmap);
p = new;
}
rcu_assign_pointer(idp->top, p);
idp->layers = layers;
idr: implement idr_preload[_end]() and idr_alloc() The current idr interface is very cumbersome. * For all allocations, two function calls - idr_pre_get() and idr_get_new*() - should be made. * idr_pre_get() doesn't guarantee that the following idr_get_new*() will not fail from memory shortage. If idr_get_new*() returns -EAGAIN, the caller is expected to retry pre_get and allocation. * idr_get_new*() can't enforce upper limit. Upper limit can only be enforced by allocating and then freeing if above limit. * idr_layer buffer is unnecessarily per-idr. Each idr ends up keeping around MAX_IDR_FREE idr_layers. The memory consumed per idr is under two pages but it makes it difficult to make idr_layer larger. This patch implements the following new set of allocation functions. * idr_preload[_end]() - Similar to radix preload but doesn't fail. The first idr_alloc() inside preload section can be treated as if it were called with @gfp_mask used for idr_preload(). * idr_alloc() - Allocate an ID w/ lower and upper limits. Takes @gfp_flags and can be used w/o preloading. When used inside preloaded section, the allocation mask of preloading can be assumed. If idr_alloc() can be called from a context which allows sufficiently relaxed @gfp_mask, it can be used by itself. If, for example, idr_alloc() is called inside spinlock protected region, preloading can be used like the following. idr_preload(GFP_KERNEL); spin_lock(lock); id = idr_alloc(idr, ptr, start, end, GFP_NOWAIT); spin_unlock(lock); idr_preload_end(); if (id < 0) error; which is much simpler and less error-prone than idr_pre_get and idr_get_new*() loop. The new interface uses per-pcu idr_layer buffer and thus the number of idr's in the system doesn't affect the amount of memory used for preloading. idr_layer_alloc() is introduced to handle idr_layer allocations for both old and new ID allocation paths. This is a bit hairy now but the new interface is expected to replace the old and the internal implementation eventually will become simpler. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:03:55 +00:00
v = sub_alloc(idp, &id, pa, gfp_mask, layer_idr);
if (v == -EAGAIN)
goto build_up;
return(v);
}
/*
* @id and @pa are from a successful allocation from idr_get_empty_slot().
* Install the user pointer @ptr and mark the slot full.
*/
static void idr_fill_slot(struct idr *idr, void *ptr, int id,
struct idr_layer **pa)
{
/* update hint used for lookup, cleared from free_layer() */
rcu_assign_pointer(idr->hint, pa[0]);
rcu_assign_pointer(pa[0]->ary[id & IDR_MASK], (struct idr_layer *)ptr);
pa[0]->count++;
idr_mark_full(pa, id);
}
int __idr_get_new_above(struct idr *idp, void *ptr, int starting_id, int *id)
{
idr: fix top layer handling Most functions in idr fail to deal with the high bits when the idr tree grows to the maximum height. * idr_get_empty_slot() stops growing idr tree once the depth reaches MAX_IDR_LEVEL - 1, which is one depth shallower than necessary to cover the whole range. The function doesn't even notice that it didn't grow the tree enough and ends up allocating the wrong ID given sufficiently high @starting_id. For example, on 64 bit, if the starting id is 0x7fffff01, idr_get_empty_slot() will grow the tree 5 layer deep, which only covers the 30 bits and then proceed to allocate as if the bit 30 wasn't specified. It ends up allocating 0x3fffff01 without the bit 30 but still returns 0x7fffff01. * __idr_remove_all() will not remove anything if the tree is fully grown. * idr_find() can't find anything if the tree is fully grown. * idr_for_each() and idr_get_next() can't iterate anything if the tree is fully grown. Fix it by introducing idr_max() which returns the maximum possible ID given the depth of tree and replacing the id limit checks in all affected places. As the idr_layer pointer array pa[] needs to be 1 larger than the maximum depth, enlarge pa[] arrays by one. While this plugs the discovered issues, the whole code base is horrible and in desparate need of rewrite. It's fragile like hell, Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:05:02 +00:00
struct idr_layer *pa[MAX_IDR_LEVEL + 1];
int rv;
idr: implement idr_preload[_end]() and idr_alloc() The current idr interface is very cumbersome. * For all allocations, two function calls - idr_pre_get() and idr_get_new*() - should be made. * idr_pre_get() doesn't guarantee that the following idr_get_new*() will not fail from memory shortage. If idr_get_new*() returns -EAGAIN, the caller is expected to retry pre_get and allocation. * idr_get_new*() can't enforce upper limit. Upper limit can only be enforced by allocating and then freeing if above limit. * idr_layer buffer is unnecessarily per-idr. Each idr ends up keeping around MAX_IDR_FREE idr_layers. The memory consumed per idr is under two pages but it makes it difficult to make idr_layer larger. This patch implements the following new set of allocation functions. * idr_preload[_end]() - Similar to radix preload but doesn't fail. The first idr_alloc() inside preload section can be treated as if it were called with @gfp_mask used for idr_preload(). * idr_alloc() - Allocate an ID w/ lower and upper limits. Takes @gfp_flags and can be used w/o preloading. When used inside preloaded section, the allocation mask of preloading can be assumed. If idr_alloc() can be called from a context which allows sufficiently relaxed @gfp_mask, it can be used by itself. If, for example, idr_alloc() is called inside spinlock protected region, preloading can be used like the following. idr_preload(GFP_KERNEL); spin_lock(lock); id = idr_alloc(idr, ptr, start, end, GFP_NOWAIT); spin_unlock(lock); idr_preload_end(); if (id < 0) error; which is much simpler and less error-prone than idr_pre_get and idr_get_new*() loop. The new interface uses per-pcu idr_layer buffer and thus the number of idr's in the system doesn't affect the amount of memory used for preloading. idr_layer_alloc() is introduced to handle idr_layer allocations for both old and new ID allocation paths. This is a bit hairy now but the new interface is expected to replace the old and the internal implementation eventually will become simpler. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:03:55 +00:00
rv = idr_get_empty_slot(idp, starting_id, pa, 0, idp);
if (rv < 0)
return rv == -ENOMEM ? -EAGAIN : rv;
idr_fill_slot(idp, ptr, rv, pa);
*id = rv;
return 0;
}
EXPORT_SYMBOL(__idr_get_new_above);
idr: implement idr_preload[_end]() and idr_alloc() The current idr interface is very cumbersome. * For all allocations, two function calls - idr_pre_get() and idr_get_new*() - should be made. * idr_pre_get() doesn't guarantee that the following idr_get_new*() will not fail from memory shortage. If idr_get_new*() returns -EAGAIN, the caller is expected to retry pre_get and allocation. * idr_get_new*() can't enforce upper limit. Upper limit can only be enforced by allocating and then freeing if above limit. * idr_layer buffer is unnecessarily per-idr. Each idr ends up keeping around MAX_IDR_FREE idr_layers. The memory consumed per idr is under two pages but it makes it difficult to make idr_layer larger. This patch implements the following new set of allocation functions. * idr_preload[_end]() - Similar to radix preload but doesn't fail. The first idr_alloc() inside preload section can be treated as if it were called with @gfp_mask used for idr_preload(). * idr_alloc() - Allocate an ID w/ lower and upper limits. Takes @gfp_flags and can be used w/o preloading. When used inside preloaded section, the allocation mask of preloading can be assumed. If idr_alloc() can be called from a context which allows sufficiently relaxed @gfp_mask, it can be used by itself. If, for example, idr_alloc() is called inside spinlock protected region, preloading can be used like the following. idr_preload(GFP_KERNEL); spin_lock(lock); id = idr_alloc(idr, ptr, start, end, GFP_NOWAIT); spin_unlock(lock); idr_preload_end(); if (id < 0) error; which is much simpler and less error-prone than idr_pre_get and idr_get_new*() loop. The new interface uses per-pcu idr_layer buffer and thus the number of idr's in the system doesn't affect the amount of memory used for preloading. idr_layer_alloc() is introduced to handle idr_layer allocations for both old and new ID allocation paths. This is a bit hairy now but the new interface is expected to replace the old and the internal implementation eventually will become simpler. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:03:55 +00:00
/**
* idr_preload - preload for idr_alloc()
* @gfp_mask: allocation mask to use for preloading
*
* Preload per-cpu layer buffer for idr_alloc(). Can only be used from
* process context and each idr_preload() invocation should be matched with
* idr_preload_end(). Note that preemption is disabled while preloaded.
*
* The first idr_alloc() in the preloaded section can be treated as if it
* were invoked with @gfp_mask used for preloading. This allows using more
* permissive allocation masks for idrs protected by spinlocks.
*
* For example, if idr_alloc() below fails, the failure can be treated as
* if idr_alloc() were called with GFP_KERNEL rather than GFP_NOWAIT.
*
* idr_preload(GFP_KERNEL);
* spin_lock(lock);
*
* id = idr_alloc(idr, ptr, start, end, GFP_NOWAIT);
*
* spin_unlock(lock);
* idr_preload_end();
* if (id < 0)
* error;
*/
void idr_preload(gfp_t gfp_mask)
{
/*
* Consuming preload buffer from non-process context breaks preload
* allocation guarantee. Disallow usage from those contexts.
*/
WARN_ON_ONCE(in_interrupt());
might_sleep_if(gfp_mask & __GFP_WAIT);
preempt_disable();
/*
* idr_alloc() is likely to succeed w/o full idr_layer buffer and
* return value from idr_alloc() needs to be checked for failure
* anyway. Silently give up if allocation fails. The caller can
* treat failures from idr_alloc() as if idr_alloc() were called
* with @gfp_mask which should be enough.
*/
while (__this_cpu_read(idr_preload_cnt) < MAX_IDR_FREE) {
struct idr_layer *new;
preempt_enable();
new = kmem_cache_zalloc(idr_layer_cache, gfp_mask);
preempt_disable();
if (!new)
break;
/* link the new one to per-cpu preload list */
new->ary[0] = __this_cpu_read(idr_preload_head);
__this_cpu_write(idr_preload_head, new);
__this_cpu_inc(idr_preload_cnt);
}
}
EXPORT_SYMBOL(idr_preload);
/**
* idr_alloc - allocate new idr entry
* @idr: the (initialized) idr
* @ptr: pointer to be associated with the new id
* @start: the minimum id (inclusive)
* @end: the maximum id (exclusive, <= 0 for max)
* @gfp_mask: memory allocation flags
*
* Allocate an id in [start, end) and associate it with @ptr. If no ID is
* available in the specified range, returns -ENOSPC. On memory allocation
* failure, returns -ENOMEM.
*
* Note that @end is treated as max when <= 0. This is to always allow
* using @start + N as @end as long as N is inside integer range.
*
* The user is responsible for exclusively synchronizing all operations
* which may modify @idr. However, read-only accesses such as idr_find()
* or iteration can be performed under RCU read lock provided the user
* destroys @ptr in RCU-safe way after removal from idr.
*/
int idr_alloc(struct idr *idr, void *ptr, int start, int end, gfp_t gfp_mask)
{
int max = end > 0 ? end - 1 : INT_MAX; /* inclusive upper limit */
idr: fix top layer handling Most functions in idr fail to deal with the high bits when the idr tree grows to the maximum height. * idr_get_empty_slot() stops growing idr tree once the depth reaches MAX_IDR_LEVEL - 1, which is one depth shallower than necessary to cover the whole range. The function doesn't even notice that it didn't grow the tree enough and ends up allocating the wrong ID given sufficiently high @starting_id. For example, on 64 bit, if the starting id is 0x7fffff01, idr_get_empty_slot() will grow the tree 5 layer deep, which only covers the 30 bits and then proceed to allocate as if the bit 30 wasn't specified. It ends up allocating 0x3fffff01 without the bit 30 but still returns 0x7fffff01. * __idr_remove_all() will not remove anything if the tree is fully grown. * idr_find() can't find anything if the tree is fully grown. * idr_for_each() and idr_get_next() can't iterate anything if the tree is fully grown. Fix it by introducing idr_max() which returns the maximum possible ID given the depth of tree and replacing the id limit checks in all affected places. As the idr_layer pointer array pa[] needs to be 1 larger than the maximum depth, enlarge pa[] arrays by one. While this plugs the discovered issues, the whole code base is horrible and in desparate need of rewrite. It's fragile like hell, Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:05:02 +00:00
struct idr_layer *pa[MAX_IDR_LEVEL + 1];
idr: implement idr_preload[_end]() and idr_alloc() The current idr interface is very cumbersome. * For all allocations, two function calls - idr_pre_get() and idr_get_new*() - should be made. * idr_pre_get() doesn't guarantee that the following idr_get_new*() will not fail from memory shortage. If idr_get_new*() returns -EAGAIN, the caller is expected to retry pre_get and allocation. * idr_get_new*() can't enforce upper limit. Upper limit can only be enforced by allocating and then freeing if above limit. * idr_layer buffer is unnecessarily per-idr. Each idr ends up keeping around MAX_IDR_FREE idr_layers. The memory consumed per idr is under two pages but it makes it difficult to make idr_layer larger. This patch implements the following new set of allocation functions. * idr_preload[_end]() - Similar to radix preload but doesn't fail. The first idr_alloc() inside preload section can be treated as if it were called with @gfp_mask used for idr_preload(). * idr_alloc() - Allocate an ID w/ lower and upper limits. Takes @gfp_flags and can be used w/o preloading. When used inside preloaded section, the allocation mask of preloading can be assumed. If idr_alloc() can be called from a context which allows sufficiently relaxed @gfp_mask, it can be used by itself. If, for example, idr_alloc() is called inside spinlock protected region, preloading can be used like the following. idr_preload(GFP_KERNEL); spin_lock(lock); id = idr_alloc(idr, ptr, start, end, GFP_NOWAIT); spin_unlock(lock); idr_preload_end(); if (id < 0) error; which is much simpler and less error-prone than idr_pre_get and idr_get_new*() loop. The new interface uses per-pcu idr_layer buffer and thus the number of idr's in the system doesn't affect the amount of memory used for preloading. idr_layer_alloc() is introduced to handle idr_layer allocations for both old and new ID allocation paths. This is a bit hairy now but the new interface is expected to replace the old and the internal implementation eventually will become simpler. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:03:55 +00:00
int id;
might_sleep_if(gfp_mask & __GFP_WAIT);
/* sanity checks */
if (WARN_ON_ONCE(start < 0))
return -EINVAL;
if (unlikely(max < start))
return -ENOSPC;
/* allocate id */
id = idr_get_empty_slot(idr, start, pa, gfp_mask, NULL);
if (unlikely(id < 0))
return id;
if (unlikely(id > max))
return -ENOSPC;
idr_fill_slot(idr, ptr, id, pa);
idr: implement idr_preload[_end]() and idr_alloc() The current idr interface is very cumbersome. * For all allocations, two function calls - idr_pre_get() and idr_get_new*() - should be made. * idr_pre_get() doesn't guarantee that the following idr_get_new*() will not fail from memory shortage. If idr_get_new*() returns -EAGAIN, the caller is expected to retry pre_get and allocation. * idr_get_new*() can't enforce upper limit. Upper limit can only be enforced by allocating and then freeing if above limit. * idr_layer buffer is unnecessarily per-idr. Each idr ends up keeping around MAX_IDR_FREE idr_layers. The memory consumed per idr is under two pages but it makes it difficult to make idr_layer larger. This patch implements the following new set of allocation functions. * idr_preload[_end]() - Similar to radix preload but doesn't fail. The first idr_alloc() inside preload section can be treated as if it were called with @gfp_mask used for idr_preload(). * idr_alloc() - Allocate an ID w/ lower and upper limits. Takes @gfp_flags and can be used w/o preloading. When used inside preloaded section, the allocation mask of preloading can be assumed. If idr_alloc() can be called from a context which allows sufficiently relaxed @gfp_mask, it can be used by itself. If, for example, idr_alloc() is called inside spinlock protected region, preloading can be used like the following. idr_preload(GFP_KERNEL); spin_lock(lock); id = idr_alloc(idr, ptr, start, end, GFP_NOWAIT); spin_unlock(lock); idr_preload_end(); if (id < 0) error; which is much simpler and less error-prone than idr_pre_get and idr_get_new*() loop. The new interface uses per-pcu idr_layer buffer and thus the number of idr's in the system doesn't affect the amount of memory used for preloading. idr_layer_alloc() is introduced to handle idr_layer allocations for both old and new ID allocation paths. This is a bit hairy now but the new interface is expected to replace the old and the internal implementation eventually will become simpler. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:03:55 +00:00
return id;
}
EXPORT_SYMBOL_GPL(idr_alloc);
idr: introduce idr_alloc_cyclic() As Tejun points out, there are several users of the IDR facility that attempt to use it in a cyclic fashion. These users are likely to see -ENOSPC errors after the counter wraps one or more times however. This patchset adds a new idr_alloc_cyclic routine and converts several of these users to it. Many of these users are in obscure parts of the kernel, and I don't have a good way to test some of them. The change is pretty straightforward though, so hopefully it won't be an issue. There is one other cyclic user of idr_alloc that I didn't touch in ipc/util.c. That one is doing some strange stuff that I didn't quite understand, but it looks like it should probably be converted later somehow. This patch: Thus spake Tejun Heo: Ooh, BTW, the cyclic allocation is broken. It's prone to -ENOSPC after the first wraparound. There are several cyclic users in the kernel and I think it probably would be best to implement cyclic support in idr. This patch does that by adding new idr_alloc_cyclic function that such users in the kernel can use. With this, there's no need for a caller to keep track of the last value used as that's now tracked internally. This should prevent the ENOSPC problems that can hit when the "last allocated" counter exceeds INT_MAX. Later patches will convert existing cyclic users to the new interface. Signed-off-by: Jeff Layton <jlayton@redhat.com> Reviewed-by: Tejun Heo <tj@kernel.org> Cc: "David S. Miller" <davem@davemloft.net> Cc: "J. Bruce Fields" <bfields@fieldses.org> Cc: Eric Paris <eparis@parisplace.org> Cc: Jack Morgenstein <jackm@dev.mellanox.co.il> Cc: John McCutchan <john@johnmccutchan.com> Cc: Neil Horman <nhorman@tuxdriver.com> Cc: Or Gerlitz <ogerlitz@mellanox.com> Cc: Robert Love <rlove@rlove.org> Cc: Roland Dreier <roland@purestorage.com> Cc: Sridhar Samudrala <sri@us.ibm.com> Cc: Steve Wise <swise@opengridcomputing.com> Cc: Tom Tucker <tom@opengridcomputing.com> Cc: Vlad Yasevich <vyasevich@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-04-29 23:21:16 +00:00
/**
* idr_alloc_cyclic - allocate new idr entry in a cyclical fashion
* @idr: the (initialized) idr
* @ptr: pointer to be associated with the new id
* @start: the minimum id (inclusive)
* @end: the maximum id (exclusive, <= 0 for max)
* @gfp_mask: memory allocation flags
*
* Essentially the same as idr_alloc, but prefers to allocate progressively
* higher ids if it can. If the "cur" counter wraps, then it will start again
* at the "start" end of the range and allocate one that has already been used.
*/
int idr_alloc_cyclic(struct idr *idr, void *ptr, int start, int end,
gfp_t gfp_mask)
{
int id;
id = idr_alloc(idr, ptr, max(start, idr->cur), end, gfp_mask);
if (id == -ENOSPC)
id = idr_alloc(idr, ptr, start, end, gfp_mask);
if (likely(id >= 0))
idr->cur = id + 1;
return id;
}
EXPORT_SYMBOL(idr_alloc_cyclic);
static void idr_remove_warning(int id)
{
printk(KERN_WARNING
"idr_remove called for id=%d which is not allocated.\n", id);
dump_stack();
}
static void sub_remove(struct idr *idp, int shift, int id)
{
struct idr_layer *p = idp->top;
idr: fix top layer handling Most functions in idr fail to deal with the high bits when the idr tree grows to the maximum height. * idr_get_empty_slot() stops growing idr tree once the depth reaches MAX_IDR_LEVEL - 1, which is one depth shallower than necessary to cover the whole range. The function doesn't even notice that it didn't grow the tree enough and ends up allocating the wrong ID given sufficiently high @starting_id. For example, on 64 bit, if the starting id is 0x7fffff01, idr_get_empty_slot() will grow the tree 5 layer deep, which only covers the 30 bits and then proceed to allocate as if the bit 30 wasn't specified. It ends up allocating 0x3fffff01 without the bit 30 but still returns 0x7fffff01. * __idr_remove_all() will not remove anything if the tree is fully grown. * idr_find() can't find anything if the tree is fully grown. * idr_for_each() and idr_get_next() can't iterate anything if the tree is fully grown. Fix it by introducing idr_max() which returns the maximum possible ID given the depth of tree and replacing the id limit checks in all affected places. As the idr_layer pointer array pa[] needs to be 1 larger than the maximum depth, enlarge pa[] arrays by one. While this plugs the discovered issues, the whole code base is horrible and in desparate need of rewrite. It's fragile like hell, Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:05:02 +00:00
struct idr_layer **pa[MAX_IDR_LEVEL + 1];
struct idr_layer ***paa = &pa[0];
struct idr_layer *to_free;
int n;
*paa = NULL;
*++paa = &idp->top;
while ((shift > 0) && p) {
n = (id >> shift) & IDR_MASK;
__clear_bit(n, p->bitmap);
*++paa = &p->ary[n];
p = p->ary[n];
shift -= IDR_BITS;
}
n = id & IDR_MASK;
if (likely(p != NULL && test_bit(n, p->bitmap))) {
__clear_bit(n, p->bitmap);
rcu_assign_pointer(p->ary[n], NULL);
to_free = NULL;
while(*paa && ! --((**paa)->count)){
if (to_free)
free_layer(idp, to_free);
to_free = **paa;
**paa-- = NULL;
}
if (!*paa)
idp->layers = 0;
if (to_free)
free_layer(idp, to_free);
} else
idr_remove_warning(id);
}
/**
* idr_remove - remove the given id and free its slot
* @idp: idr handle
* @id: unique key
*/
void idr_remove(struct idr *idp, int id)
{
struct idr_layer *p;
struct idr_layer *to_free;
if (id < 0)
idr: remove MAX_IDR_MASK and move left MAX_IDR_* into idr.c MAX_IDR_MASK is another weirdness in the idr interface. As idr covers whole positive integer range, it's defined as 0x7fffffff or INT_MAX. Its usage in idr_find(), idr_replace() and idr_remove() is bizarre. They basically mask off the sign bit and operate on the rest, so if the caller, by accident, passes in a negative number, the sign bit will be masked off and the remaining part will be used as if that was the input, which is worse than crashing. The constant is visible in idr.h and there are several users in the kernel. * drivers/i2c/i2c-core.c:i2c_add_numbered_adapter() Basically used to test if adap->nr is a negative number which isn't -1 and returns -EINVAL if so. idr_alloc() already has negative @start checking (w/ WARN_ON_ONCE), so this can go away. * drivers/infiniband/core/cm.c:cm_alloc_id() drivers/infiniband/hw/mlx4/cm.c:id_map_alloc() Used to wrap cyclic @start. Can be replaced with max(next, 0). Note that this type of cyclic allocation using idr is buggy. These are prone to spurious -ENOSPC failure after the first wraparound. * fs/super.c:get_anon_bdev() The ID allocated from ida is masked off before being tested whether it's inside valid range. ida allocated ID can never be a negative number and the masking is unnecessary. Update idr_*() functions to fail with -EINVAL when negative @id is specified and update other MAX_IDR_MASK users as described above. This leaves MAX_IDR_MASK without any user, remove it and relocate other MAX_IDR_* constants to lib/idr.c. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Roland Dreier <roland@kernel.org> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: "Marciniszyn, Mike" <mike.marciniszyn@intel.com> Cc: Jack Morgenstein <jackm@dev.mellanox.co.il> Cc: Or Gerlitz <ogerlitz@mellanox.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Acked-by: Wolfram Sang <wolfram@the-dreams.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:05:04 +00:00
return;
sub_remove(idp, (idp->layers - 1) * IDR_BITS, id);
if (idp->top && idp->top->count == 1 && (idp->layers > 1) &&
idp->top->ary[0]) {
/*
* Single child at leftmost slot: we can shrink the tree.
* This level is not needed anymore since when layers are
* inserted, they are inserted at the top of the existing
* tree.
*/
to_free = idp->top;
p = idp->top->ary[0];
rcu_assign_pointer(idp->top, p);
--idp->layers;
to_free->count = 0;
bitmap_clear(to_free->bitmap, 0, IDR_SIZE);
free_layer(idp, to_free);
}
while (idp->id_free_cnt >= MAX_IDR_FREE) {
p = get_from_free_list(idp);
/*
* Note: we don't call the rcu callback here, since the only
* layers that fall into the freelist are those that have been
* preallocated.
*/
kmem_cache_free(idr_layer_cache, p);
}
return;
}
EXPORT_SYMBOL(idr_remove);
void __idr_remove_all(struct idr *idp)
{
int n, id, max;
idr: fix backtrack logic in idr_remove_all Currently idr_remove_all will fail with a use after free error if idr::layers is bigger than 2, which on 32 bit systems corresponds to items more than 1024. This is due to stepping back too many levels during backtracking. For simplicity let's assume that IDR_BITS=1 -> we have 2 nodes at each level below the root node and each leaf node stores two IDs. (In reality for 32 bit systems IDR_BITS=5, with 32 nodes at each sub-root level and 32 IDs in each leaf node). The sequence of freeing the nodes at the moment is as follows: layer 1 -> a(7) 2 -> b(3) c(5) 3 -> d(1) e(2) f(4) g(6) Until step 4 things go fine, but then node c is freed, whereas node g should be freed first. Since node c contains the pointer to node g we'll have a use after free error at step 6. How many levels we step back after visiting the leaf nodes is currently determined by the msb of the id we are currently visiting: Step 1. node d with IDs 0,1 is freed, current ID is advanced to 2. msb of the current ID bit 1. This means we need to step back 1 level to node b and take the next sibling, node e. 2-3. node e with IDs 2,3 is freed, current ID is 4, msb is bit 2. This means we need to step back 2 levels to node a, freeing node b on the way. 4-5. node f with IDs 4,5 is freed, current ID is 6, msb is still bit 2. This means we again need to step back 2 levels to node a and free c on the way. 6. We should visit node g, but its pointer is not available as node c was freed. The fix changes how we determine the number of levels to step back. Instead of deducting this merely from the msb of the current ID, we should really check if advancing the ID causes an overflow to a bit position corresponding to a given layer. In the above example overflow from bit 0 to bit 1 should mean stepping back 1 level. Overflow from bit 1 to bit 2 should mean stepping back 2 levels and so on. The fix was tested with IDs up to 1 << 20, which corresponds to 4 layers on 32 bit systems. Signed-off-by: Imre Deak <imre.deak@nokia.com> Reviewed-by: Tejun Heo <tj@kernel.org> Cc: Eric Paris <eparis@redhat.com> Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> Cc: <stable@kernel.org> [2.6.34.1] Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-05-26 21:43:38 +00:00
int bt_mask;
struct idr_layer *p;
idr: fix top layer handling Most functions in idr fail to deal with the high bits when the idr tree grows to the maximum height. * idr_get_empty_slot() stops growing idr tree once the depth reaches MAX_IDR_LEVEL - 1, which is one depth shallower than necessary to cover the whole range. The function doesn't even notice that it didn't grow the tree enough and ends up allocating the wrong ID given sufficiently high @starting_id. For example, on 64 bit, if the starting id is 0x7fffff01, idr_get_empty_slot() will grow the tree 5 layer deep, which only covers the 30 bits and then proceed to allocate as if the bit 30 wasn't specified. It ends up allocating 0x3fffff01 without the bit 30 but still returns 0x7fffff01. * __idr_remove_all() will not remove anything if the tree is fully grown. * idr_find() can't find anything if the tree is fully grown. * idr_for_each() and idr_get_next() can't iterate anything if the tree is fully grown. Fix it by introducing idr_max() which returns the maximum possible ID given the depth of tree and replacing the id limit checks in all affected places. As the idr_layer pointer array pa[] needs to be 1 larger than the maximum depth, enlarge pa[] arrays by one. While this plugs the discovered issues, the whole code base is horrible and in desparate need of rewrite. It's fragile like hell, Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:05:02 +00:00
struct idr_layer *pa[MAX_IDR_LEVEL + 1];
struct idr_layer **paa = &pa[0];
n = idp->layers * IDR_BITS;
p = idp->top;
rcu_assign_pointer(idp->top, NULL);
idr: fix top layer handling Most functions in idr fail to deal with the high bits when the idr tree grows to the maximum height. * idr_get_empty_slot() stops growing idr tree once the depth reaches MAX_IDR_LEVEL - 1, which is one depth shallower than necessary to cover the whole range. The function doesn't even notice that it didn't grow the tree enough and ends up allocating the wrong ID given sufficiently high @starting_id. For example, on 64 bit, if the starting id is 0x7fffff01, idr_get_empty_slot() will grow the tree 5 layer deep, which only covers the 30 bits and then proceed to allocate as if the bit 30 wasn't specified. It ends up allocating 0x3fffff01 without the bit 30 but still returns 0x7fffff01. * __idr_remove_all() will not remove anything if the tree is fully grown. * idr_find() can't find anything if the tree is fully grown. * idr_for_each() and idr_get_next() can't iterate anything if the tree is fully grown. Fix it by introducing idr_max() which returns the maximum possible ID given the depth of tree and replacing the id limit checks in all affected places. As the idr_layer pointer array pa[] needs to be 1 larger than the maximum depth, enlarge pa[] arrays by one. While this plugs the discovered issues, the whole code base is horrible and in desparate need of rewrite. It's fragile like hell, Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:05:02 +00:00
max = idr_max(idp->layers);
id = 0;
idr: fix top layer handling Most functions in idr fail to deal with the high bits when the idr tree grows to the maximum height. * idr_get_empty_slot() stops growing idr tree once the depth reaches MAX_IDR_LEVEL - 1, which is one depth shallower than necessary to cover the whole range. The function doesn't even notice that it didn't grow the tree enough and ends up allocating the wrong ID given sufficiently high @starting_id. For example, on 64 bit, if the starting id is 0x7fffff01, idr_get_empty_slot() will grow the tree 5 layer deep, which only covers the 30 bits and then proceed to allocate as if the bit 30 wasn't specified. It ends up allocating 0x3fffff01 without the bit 30 but still returns 0x7fffff01. * __idr_remove_all() will not remove anything if the tree is fully grown. * idr_find() can't find anything if the tree is fully grown. * idr_for_each() and idr_get_next() can't iterate anything if the tree is fully grown. Fix it by introducing idr_max() which returns the maximum possible ID given the depth of tree and replacing the id limit checks in all affected places. As the idr_layer pointer array pa[] needs to be 1 larger than the maximum depth, enlarge pa[] arrays by one. While this plugs the discovered issues, the whole code base is horrible and in desparate need of rewrite. It's fragile like hell, Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:05:02 +00:00
while (id >= 0 && id <= max) {
while (n > IDR_BITS && p) {
n -= IDR_BITS;
*paa++ = p;
p = p->ary[(id >> n) & IDR_MASK];
}
idr: fix backtrack logic in idr_remove_all Currently idr_remove_all will fail with a use after free error if idr::layers is bigger than 2, which on 32 bit systems corresponds to items more than 1024. This is due to stepping back too many levels during backtracking. For simplicity let's assume that IDR_BITS=1 -> we have 2 nodes at each level below the root node and each leaf node stores two IDs. (In reality for 32 bit systems IDR_BITS=5, with 32 nodes at each sub-root level and 32 IDs in each leaf node). The sequence of freeing the nodes at the moment is as follows: layer 1 -> a(7) 2 -> b(3) c(5) 3 -> d(1) e(2) f(4) g(6) Until step 4 things go fine, but then node c is freed, whereas node g should be freed first. Since node c contains the pointer to node g we'll have a use after free error at step 6. How many levels we step back after visiting the leaf nodes is currently determined by the msb of the id we are currently visiting: Step 1. node d with IDs 0,1 is freed, current ID is advanced to 2. msb of the current ID bit 1. This means we need to step back 1 level to node b and take the next sibling, node e. 2-3. node e with IDs 2,3 is freed, current ID is 4, msb is bit 2. This means we need to step back 2 levels to node a, freeing node b on the way. 4-5. node f with IDs 4,5 is freed, current ID is 6, msb is still bit 2. This means we again need to step back 2 levels to node a and free c on the way. 6. We should visit node g, but its pointer is not available as node c was freed. The fix changes how we determine the number of levels to step back. Instead of deducting this merely from the msb of the current ID, we should really check if advancing the ID causes an overflow to a bit position corresponding to a given layer. In the above example overflow from bit 0 to bit 1 should mean stepping back 1 level. Overflow from bit 1 to bit 2 should mean stepping back 2 levels and so on. The fix was tested with IDs up to 1 << 20, which corresponds to 4 layers on 32 bit systems. Signed-off-by: Imre Deak <imre.deak@nokia.com> Reviewed-by: Tejun Heo <tj@kernel.org> Cc: Eric Paris <eparis@redhat.com> Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> Cc: <stable@kernel.org> [2.6.34.1] Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-05-26 21:43:38 +00:00
bt_mask = id;
id += 1 << n;
idr: fix backtrack logic in idr_remove_all Currently idr_remove_all will fail with a use after free error if idr::layers is bigger than 2, which on 32 bit systems corresponds to items more than 1024. This is due to stepping back too many levels during backtracking. For simplicity let's assume that IDR_BITS=1 -> we have 2 nodes at each level below the root node and each leaf node stores two IDs. (In reality for 32 bit systems IDR_BITS=5, with 32 nodes at each sub-root level and 32 IDs in each leaf node). The sequence of freeing the nodes at the moment is as follows: layer 1 -> a(7) 2 -> b(3) c(5) 3 -> d(1) e(2) f(4) g(6) Until step 4 things go fine, but then node c is freed, whereas node g should be freed first. Since node c contains the pointer to node g we'll have a use after free error at step 6. How many levels we step back after visiting the leaf nodes is currently determined by the msb of the id we are currently visiting: Step 1. node d with IDs 0,1 is freed, current ID is advanced to 2. msb of the current ID bit 1. This means we need to step back 1 level to node b and take the next sibling, node e. 2-3. node e with IDs 2,3 is freed, current ID is 4, msb is bit 2. This means we need to step back 2 levels to node a, freeing node b on the way. 4-5. node f with IDs 4,5 is freed, current ID is 6, msb is still bit 2. This means we again need to step back 2 levels to node a and free c on the way. 6. We should visit node g, but its pointer is not available as node c was freed. The fix changes how we determine the number of levels to step back. Instead of deducting this merely from the msb of the current ID, we should really check if advancing the ID causes an overflow to a bit position corresponding to a given layer. In the above example overflow from bit 0 to bit 1 should mean stepping back 1 level. Overflow from bit 1 to bit 2 should mean stepping back 2 levels and so on. The fix was tested with IDs up to 1 << 20, which corresponds to 4 layers on 32 bit systems. Signed-off-by: Imre Deak <imre.deak@nokia.com> Reviewed-by: Tejun Heo <tj@kernel.org> Cc: Eric Paris <eparis@redhat.com> Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> Cc: <stable@kernel.org> [2.6.34.1] Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-05-26 21:43:38 +00:00
/* Get the highest bit that the above add changed from 0->1. */
while (n < fls(id ^ bt_mask)) {
if (p)
free_layer(idp, p);
n += IDR_BITS;
p = *--paa;
}
}
idp->layers = 0;
}
EXPORT_SYMBOL(__idr_remove_all);
/**
* idr_destroy - release all cached layers within an idr tree
* @idp: idr handle
*
* Free all id mappings and all idp_layers. After this function, @idp is
* completely unused and can be freed / recycled. The caller is
* responsible for ensuring that no one else accesses @idp during or after
* idr_destroy().
*
* A typical clean-up sequence for objects stored in an idr tree will use
* idr_for_each() to free all objects, if necessay, then idr_destroy() to
* free up the id mappings and cached idr_layers.
*/
void idr_destroy(struct idr *idp)
{
__idr_remove_all(idp);
while (idp->id_free_cnt) {
struct idr_layer *p = get_from_free_list(idp);
kmem_cache_free(idr_layer_cache, p);
}
}
EXPORT_SYMBOL(idr_destroy);
void *idr_find_slowpath(struct idr *idp, int id)
{
int n;
struct idr_layer *p;
if (id < 0)
idr: remove MAX_IDR_MASK and move left MAX_IDR_* into idr.c MAX_IDR_MASK is another weirdness in the idr interface. As idr covers whole positive integer range, it's defined as 0x7fffffff or INT_MAX. Its usage in idr_find(), idr_replace() and idr_remove() is bizarre. They basically mask off the sign bit and operate on the rest, so if the caller, by accident, passes in a negative number, the sign bit will be masked off and the remaining part will be used as if that was the input, which is worse than crashing. The constant is visible in idr.h and there are several users in the kernel. * drivers/i2c/i2c-core.c:i2c_add_numbered_adapter() Basically used to test if adap->nr is a negative number which isn't -1 and returns -EINVAL if so. idr_alloc() already has negative @start checking (w/ WARN_ON_ONCE), so this can go away. * drivers/infiniband/core/cm.c:cm_alloc_id() drivers/infiniband/hw/mlx4/cm.c:id_map_alloc() Used to wrap cyclic @start. Can be replaced with max(next, 0). Note that this type of cyclic allocation using idr is buggy. These are prone to spurious -ENOSPC failure after the first wraparound. * fs/super.c:get_anon_bdev() The ID allocated from ida is masked off before being tested whether it's inside valid range. ida allocated ID can never be a negative number and the masking is unnecessary. Update idr_*() functions to fail with -EINVAL when negative @id is specified and update other MAX_IDR_MASK users as described above. This leaves MAX_IDR_MASK without any user, remove it and relocate other MAX_IDR_* constants to lib/idr.c. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Roland Dreier <roland@kernel.org> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: "Marciniszyn, Mike" <mike.marciniszyn@intel.com> Cc: Jack Morgenstein <jackm@dev.mellanox.co.il> Cc: Or Gerlitz <ogerlitz@mellanox.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Acked-by: Wolfram Sang <wolfram@the-dreams.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:05:04 +00:00
return NULL;
p = rcu_dereference_raw(idp->top);
if (!p)
return NULL;
n = (p->layer+1) * IDR_BITS;
idr: fix top layer handling Most functions in idr fail to deal with the high bits when the idr tree grows to the maximum height. * idr_get_empty_slot() stops growing idr tree once the depth reaches MAX_IDR_LEVEL - 1, which is one depth shallower than necessary to cover the whole range. The function doesn't even notice that it didn't grow the tree enough and ends up allocating the wrong ID given sufficiently high @starting_id. For example, on 64 bit, if the starting id is 0x7fffff01, idr_get_empty_slot() will grow the tree 5 layer deep, which only covers the 30 bits and then proceed to allocate as if the bit 30 wasn't specified. It ends up allocating 0x3fffff01 without the bit 30 but still returns 0x7fffff01. * __idr_remove_all() will not remove anything if the tree is fully grown. * idr_find() can't find anything if the tree is fully grown. * idr_for_each() and idr_get_next() can't iterate anything if the tree is fully grown. Fix it by introducing idr_max() which returns the maximum possible ID given the depth of tree and replacing the id limit checks in all affected places. As the idr_layer pointer array pa[] needs to be 1 larger than the maximum depth, enlarge pa[] arrays by one. While this plugs the discovered issues, the whole code base is horrible and in desparate need of rewrite. It's fragile like hell, Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:05:02 +00:00
if (id > idr_max(p->layer + 1))
return NULL;
BUG_ON(n == 0);
while (n > 0 && p) {
n -= IDR_BITS;
BUG_ON(n != p->layer*IDR_BITS);
p = rcu_dereference_raw(p->ary[(id >> n) & IDR_MASK]);
}
return((void *)p);
}
EXPORT_SYMBOL(idr_find_slowpath);
/**
* idr_for_each - iterate through all stored pointers
* @idp: idr handle
* @fn: function to be called for each pointer
* @data: data passed back to callback function
*
* Iterate over the pointers registered with the given idr. The
* callback function will be called for each pointer currently
* registered, passing the id, the pointer and the data pointer passed
* to this function. It is not safe to modify the idr tree while in
* the callback, so functions such as idr_get_new and idr_remove are
* not allowed.
*
* We check the return of @fn each time. If it returns anything other
* than %0, we break out and return that value.
*
* The caller must serialize idr_for_each() vs idr_get_new() and idr_remove().
*/
int idr_for_each(struct idr *idp,
int (*fn)(int id, void *p, void *data), void *data)
{
int n, id, max, error = 0;
struct idr_layer *p;
idr: fix top layer handling Most functions in idr fail to deal with the high bits when the idr tree grows to the maximum height. * idr_get_empty_slot() stops growing idr tree once the depth reaches MAX_IDR_LEVEL - 1, which is one depth shallower than necessary to cover the whole range. The function doesn't even notice that it didn't grow the tree enough and ends up allocating the wrong ID given sufficiently high @starting_id. For example, on 64 bit, if the starting id is 0x7fffff01, idr_get_empty_slot() will grow the tree 5 layer deep, which only covers the 30 bits and then proceed to allocate as if the bit 30 wasn't specified. It ends up allocating 0x3fffff01 without the bit 30 but still returns 0x7fffff01. * __idr_remove_all() will not remove anything if the tree is fully grown. * idr_find() can't find anything if the tree is fully grown. * idr_for_each() and idr_get_next() can't iterate anything if the tree is fully grown. Fix it by introducing idr_max() which returns the maximum possible ID given the depth of tree and replacing the id limit checks in all affected places. As the idr_layer pointer array pa[] needs to be 1 larger than the maximum depth, enlarge pa[] arrays by one. While this plugs the discovered issues, the whole code base is horrible and in desparate need of rewrite. It's fragile like hell, Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:05:02 +00:00
struct idr_layer *pa[MAX_IDR_LEVEL + 1];
struct idr_layer **paa = &pa[0];
n = idp->layers * IDR_BITS;
p = rcu_dereference_raw(idp->top);
idr: fix top layer handling Most functions in idr fail to deal with the high bits when the idr tree grows to the maximum height. * idr_get_empty_slot() stops growing idr tree once the depth reaches MAX_IDR_LEVEL - 1, which is one depth shallower than necessary to cover the whole range. The function doesn't even notice that it didn't grow the tree enough and ends up allocating the wrong ID given sufficiently high @starting_id. For example, on 64 bit, if the starting id is 0x7fffff01, idr_get_empty_slot() will grow the tree 5 layer deep, which only covers the 30 bits and then proceed to allocate as if the bit 30 wasn't specified. It ends up allocating 0x3fffff01 without the bit 30 but still returns 0x7fffff01. * __idr_remove_all() will not remove anything if the tree is fully grown. * idr_find() can't find anything if the tree is fully grown. * idr_for_each() and idr_get_next() can't iterate anything if the tree is fully grown. Fix it by introducing idr_max() which returns the maximum possible ID given the depth of tree and replacing the id limit checks in all affected places. As the idr_layer pointer array pa[] needs to be 1 larger than the maximum depth, enlarge pa[] arrays by one. While this plugs the discovered issues, the whole code base is horrible and in desparate need of rewrite. It's fragile like hell, Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:05:02 +00:00
max = idr_max(idp->layers);
id = 0;
idr: fix top layer handling Most functions in idr fail to deal with the high bits when the idr tree grows to the maximum height. * idr_get_empty_slot() stops growing idr tree once the depth reaches MAX_IDR_LEVEL - 1, which is one depth shallower than necessary to cover the whole range. The function doesn't even notice that it didn't grow the tree enough and ends up allocating the wrong ID given sufficiently high @starting_id. For example, on 64 bit, if the starting id is 0x7fffff01, idr_get_empty_slot() will grow the tree 5 layer deep, which only covers the 30 bits and then proceed to allocate as if the bit 30 wasn't specified. It ends up allocating 0x3fffff01 without the bit 30 but still returns 0x7fffff01. * __idr_remove_all() will not remove anything if the tree is fully grown. * idr_find() can't find anything if the tree is fully grown. * idr_for_each() and idr_get_next() can't iterate anything if the tree is fully grown. Fix it by introducing idr_max() which returns the maximum possible ID given the depth of tree and replacing the id limit checks in all affected places. As the idr_layer pointer array pa[] needs to be 1 larger than the maximum depth, enlarge pa[] arrays by one. While this plugs the discovered issues, the whole code base is horrible and in desparate need of rewrite. It's fragile like hell, Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:05:02 +00:00
while (id >= 0 && id <= max) {
while (n > 0 && p) {
n -= IDR_BITS;
*paa++ = p;
p = rcu_dereference_raw(p->ary[(id >> n) & IDR_MASK]);
}
if (p) {
error = fn(id, (void *)p, data);
if (error)
break;
}
id += 1 << n;
while (n < fls(id)) {
n += IDR_BITS;
p = *--paa;
}
}
return error;
}
EXPORT_SYMBOL(idr_for_each);
cgroup: CSS ID support Patch for Per-CSS(Cgroup Subsys State) ID and private hierarchy code. This patch attaches unique ID to each css and provides following. - css_lookup(subsys, id) returns pointer to struct cgroup_subysys_state of id. - css_get_next(subsys, id, rootid, depth, foundid) returns the next css under "root" by scanning When cgroup_subsys->use_id is set, an id for css is maintained. The cgroup framework only parepares - css_id of root css for subsys - id is automatically attached at creation of css. - id is *not* freed automatically. Because the cgroup framework don't know lifetime of cgroup_subsys_state. free_css_id() function is provided. This must be called by subsys. There are several reasons to develop this. - Saving space .... For example, memcg's swap_cgroup is array of pointers to cgroup. But it is not necessary to be very fast. By replacing pointers(8bytes per ent) to ID (2byes per ent), we can reduce much amount of memory usage. - Scanning without lock. CSS_ID provides "scan id under this ROOT" function. By this, scanning css under root can be written without locks. ex) do { rcu_read_lock(); next = cgroup_get_next(subsys, id, root, &found); /* check sanity of next here */ css_tryget(); rcu_read_unlock(); id = found + 1 } while(...) Characteristics: - Each css has unique ID under subsys. - Lifetime of ID is controlled by subsys. - css ID contains "ID" and "Depth in hierarchy" and stack of hierarchy - Allowed ID is 1-65535, ID 0 is UNUSED ID. Design Choices: - scan-by-ID v.s. scan-by-tree-walk. As /proc's pid scan does, scan-by-ID is robust when scanning is done by following kind of routine. scan -> rest a while(release a lock) -> conitunue from interrupted memcg's hierarchical reclaim does this. - When subsys->use_id is set, # of css in the system is limited to 65535. [bharata@linux.vnet.ibm.com: remove rcu_read_lock() from css_get_next()] Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Acked-by: Paul Menage <menage@google.com> Cc: Li Zefan <lizf@cn.fujitsu.com> Cc: Balbir Singh <balbir@in.ibm.com> Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp> Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-04-02 23:57:25 +00:00
/**
* idr_get_next - lookup next object of id to given id.
* @idp: idr handle
* @nextidp: pointer to lookup key
cgroup: CSS ID support Patch for Per-CSS(Cgroup Subsys State) ID and private hierarchy code. This patch attaches unique ID to each css and provides following. - css_lookup(subsys, id) returns pointer to struct cgroup_subysys_state of id. - css_get_next(subsys, id, rootid, depth, foundid) returns the next css under "root" by scanning When cgroup_subsys->use_id is set, an id for css is maintained. The cgroup framework only parepares - css_id of root css for subsys - id is automatically attached at creation of css. - id is *not* freed automatically. Because the cgroup framework don't know lifetime of cgroup_subsys_state. free_css_id() function is provided. This must be called by subsys. There are several reasons to develop this. - Saving space .... For example, memcg's swap_cgroup is array of pointers to cgroup. But it is not necessary to be very fast. By replacing pointers(8bytes per ent) to ID (2byes per ent), we can reduce much amount of memory usage. - Scanning without lock. CSS_ID provides "scan id under this ROOT" function. By this, scanning css under root can be written without locks. ex) do { rcu_read_lock(); next = cgroup_get_next(subsys, id, root, &found); /* check sanity of next here */ css_tryget(); rcu_read_unlock(); id = found + 1 } while(...) Characteristics: - Each css has unique ID under subsys. - Lifetime of ID is controlled by subsys. - css ID contains "ID" and "Depth in hierarchy" and stack of hierarchy - Allowed ID is 1-65535, ID 0 is UNUSED ID. Design Choices: - scan-by-ID v.s. scan-by-tree-walk. As /proc's pid scan does, scan-by-ID is robust when scanning is done by following kind of routine. scan -> rest a while(release a lock) -> conitunue from interrupted memcg's hierarchical reclaim does this. - When subsys->use_id is set, # of css in the system is limited to 65535. [bharata@linux.vnet.ibm.com: remove rcu_read_lock() from css_get_next()] Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Acked-by: Paul Menage <menage@google.com> Cc: Li Zefan <lizf@cn.fujitsu.com> Cc: Balbir Singh <balbir@in.ibm.com> Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp> Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-04-02 23:57:25 +00:00
*
* Returns pointer to registered object with id, which is next number to
* given id. After being looked up, *@nextidp will be updated for the next
* iteration.
*
* This function can be called under rcu_read_lock(), given that the leaf
* pointers lifetimes are correctly managed.
cgroup: CSS ID support Patch for Per-CSS(Cgroup Subsys State) ID and private hierarchy code. This patch attaches unique ID to each css and provides following. - css_lookup(subsys, id) returns pointer to struct cgroup_subysys_state of id. - css_get_next(subsys, id, rootid, depth, foundid) returns the next css under "root" by scanning When cgroup_subsys->use_id is set, an id for css is maintained. The cgroup framework only parepares - css_id of root css for subsys - id is automatically attached at creation of css. - id is *not* freed automatically. Because the cgroup framework don't know lifetime of cgroup_subsys_state. free_css_id() function is provided. This must be called by subsys. There are several reasons to develop this. - Saving space .... For example, memcg's swap_cgroup is array of pointers to cgroup. But it is not necessary to be very fast. By replacing pointers(8bytes per ent) to ID (2byes per ent), we can reduce much amount of memory usage. - Scanning without lock. CSS_ID provides "scan id under this ROOT" function. By this, scanning css under root can be written without locks. ex) do { rcu_read_lock(); next = cgroup_get_next(subsys, id, root, &found); /* check sanity of next here */ css_tryget(); rcu_read_unlock(); id = found + 1 } while(...) Characteristics: - Each css has unique ID under subsys. - Lifetime of ID is controlled by subsys. - css ID contains "ID" and "Depth in hierarchy" and stack of hierarchy - Allowed ID is 1-65535, ID 0 is UNUSED ID. Design Choices: - scan-by-ID v.s. scan-by-tree-walk. As /proc's pid scan does, scan-by-ID is robust when scanning is done by following kind of routine. scan -> rest a while(release a lock) -> conitunue from interrupted memcg's hierarchical reclaim does this. - When subsys->use_id is set, # of css in the system is limited to 65535. [bharata@linux.vnet.ibm.com: remove rcu_read_lock() from css_get_next()] Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Acked-by: Paul Menage <menage@google.com> Cc: Li Zefan <lizf@cn.fujitsu.com> Cc: Balbir Singh <balbir@in.ibm.com> Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp> Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-04-02 23:57:25 +00:00
*/
void *idr_get_next(struct idr *idp, int *nextidp)
{
idr: fix top layer handling Most functions in idr fail to deal with the high bits when the idr tree grows to the maximum height. * idr_get_empty_slot() stops growing idr tree once the depth reaches MAX_IDR_LEVEL - 1, which is one depth shallower than necessary to cover the whole range. The function doesn't even notice that it didn't grow the tree enough and ends up allocating the wrong ID given sufficiently high @starting_id. For example, on 64 bit, if the starting id is 0x7fffff01, idr_get_empty_slot() will grow the tree 5 layer deep, which only covers the 30 bits and then proceed to allocate as if the bit 30 wasn't specified. It ends up allocating 0x3fffff01 without the bit 30 but still returns 0x7fffff01. * __idr_remove_all() will not remove anything if the tree is fully grown. * idr_find() can't find anything if the tree is fully grown. * idr_for_each() and idr_get_next() can't iterate anything if the tree is fully grown. Fix it by introducing idr_max() which returns the maximum possible ID given the depth of tree and replacing the id limit checks in all affected places. As the idr_layer pointer array pa[] needs to be 1 larger than the maximum depth, enlarge pa[] arrays by one. While this plugs the discovered issues, the whole code base is horrible and in desparate need of rewrite. It's fragile like hell, Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:05:02 +00:00
struct idr_layer *p, *pa[MAX_IDR_LEVEL + 1];
cgroup: CSS ID support Patch for Per-CSS(Cgroup Subsys State) ID and private hierarchy code. This patch attaches unique ID to each css and provides following. - css_lookup(subsys, id) returns pointer to struct cgroup_subysys_state of id. - css_get_next(subsys, id, rootid, depth, foundid) returns the next css under "root" by scanning When cgroup_subsys->use_id is set, an id for css is maintained. The cgroup framework only parepares - css_id of root css for subsys - id is automatically attached at creation of css. - id is *not* freed automatically. Because the cgroup framework don't know lifetime of cgroup_subsys_state. free_css_id() function is provided. This must be called by subsys. There are several reasons to develop this. - Saving space .... For example, memcg's swap_cgroup is array of pointers to cgroup. But it is not necessary to be very fast. By replacing pointers(8bytes per ent) to ID (2byes per ent), we can reduce much amount of memory usage. - Scanning without lock. CSS_ID provides "scan id under this ROOT" function. By this, scanning css under root can be written without locks. ex) do { rcu_read_lock(); next = cgroup_get_next(subsys, id, root, &found); /* check sanity of next here */ css_tryget(); rcu_read_unlock(); id = found + 1 } while(...) Characteristics: - Each css has unique ID under subsys. - Lifetime of ID is controlled by subsys. - css ID contains "ID" and "Depth in hierarchy" and stack of hierarchy - Allowed ID is 1-65535, ID 0 is UNUSED ID. Design Choices: - scan-by-ID v.s. scan-by-tree-walk. As /proc's pid scan does, scan-by-ID is robust when scanning is done by following kind of routine. scan -> rest a while(release a lock) -> conitunue from interrupted memcg's hierarchical reclaim does this. - When subsys->use_id is set, # of css in the system is limited to 65535. [bharata@linux.vnet.ibm.com: remove rcu_read_lock() from css_get_next()] Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Acked-by: Paul Menage <menage@google.com> Cc: Li Zefan <lizf@cn.fujitsu.com> Cc: Balbir Singh <balbir@in.ibm.com> Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp> Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-04-02 23:57:25 +00:00
struct idr_layer **paa = &pa[0];
int id = *nextidp;
int n, max;
/* find first ent */
p = rcu_dereference_raw(idp->top);
cgroup: CSS ID support Patch for Per-CSS(Cgroup Subsys State) ID and private hierarchy code. This patch attaches unique ID to each css and provides following. - css_lookup(subsys, id) returns pointer to struct cgroup_subysys_state of id. - css_get_next(subsys, id, rootid, depth, foundid) returns the next css under "root" by scanning When cgroup_subsys->use_id is set, an id for css is maintained. The cgroup framework only parepares - css_id of root css for subsys - id is automatically attached at creation of css. - id is *not* freed automatically. Because the cgroup framework don't know lifetime of cgroup_subsys_state. free_css_id() function is provided. This must be called by subsys. There are several reasons to develop this. - Saving space .... For example, memcg's swap_cgroup is array of pointers to cgroup. But it is not necessary to be very fast. By replacing pointers(8bytes per ent) to ID (2byes per ent), we can reduce much amount of memory usage. - Scanning without lock. CSS_ID provides "scan id under this ROOT" function. By this, scanning css under root can be written without locks. ex) do { rcu_read_lock(); next = cgroup_get_next(subsys, id, root, &found); /* check sanity of next here */ css_tryget(); rcu_read_unlock(); id = found + 1 } while(...) Characteristics: - Each css has unique ID under subsys. - Lifetime of ID is controlled by subsys. - css ID contains "ID" and "Depth in hierarchy" and stack of hierarchy - Allowed ID is 1-65535, ID 0 is UNUSED ID. Design Choices: - scan-by-ID v.s. scan-by-tree-walk. As /proc's pid scan does, scan-by-ID is robust when scanning is done by following kind of routine. scan -> rest a while(release a lock) -> conitunue from interrupted memcg's hierarchical reclaim does this. - When subsys->use_id is set, # of css in the system is limited to 65535. [bharata@linux.vnet.ibm.com: remove rcu_read_lock() from css_get_next()] Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Acked-by: Paul Menage <menage@google.com> Cc: Li Zefan <lizf@cn.fujitsu.com> Cc: Balbir Singh <balbir@in.ibm.com> Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp> Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-04-02 23:57:25 +00:00
if (!p)
return NULL;
n = (p->layer + 1) * IDR_BITS;
idr: fix top layer handling Most functions in idr fail to deal with the high bits when the idr tree grows to the maximum height. * idr_get_empty_slot() stops growing idr tree once the depth reaches MAX_IDR_LEVEL - 1, which is one depth shallower than necessary to cover the whole range. The function doesn't even notice that it didn't grow the tree enough and ends up allocating the wrong ID given sufficiently high @starting_id. For example, on 64 bit, if the starting id is 0x7fffff01, idr_get_empty_slot() will grow the tree 5 layer deep, which only covers the 30 bits and then proceed to allocate as if the bit 30 wasn't specified. It ends up allocating 0x3fffff01 without the bit 30 but still returns 0x7fffff01. * __idr_remove_all() will not remove anything if the tree is fully grown. * idr_find() can't find anything if the tree is fully grown. * idr_for_each() and idr_get_next() can't iterate anything if the tree is fully grown. Fix it by introducing idr_max() which returns the maximum possible ID given the depth of tree and replacing the id limit checks in all affected places. As the idr_layer pointer array pa[] needs to be 1 larger than the maximum depth, enlarge pa[] arrays by one. While this plugs the discovered issues, the whole code base is horrible and in desparate need of rewrite. It's fragile like hell, Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:05:02 +00:00
max = idr_max(p->layer + 1);
cgroup: CSS ID support Patch for Per-CSS(Cgroup Subsys State) ID and private hierarchy code. This patch attaches unique ID to each css and provides following. - css_lookup(subsys, id) returns pointer to struct cgroup_subysys_state of id. - css_get_next(subsys, id, rootid, depth, foundid) returns the next css under "root" by scanning When cgroup_subsys->use_id is set, an id for css is maintained. The cgroup framework only parepares - css_id of root css for subsys - id is automatically attached at creation of css. - id is *not* freed automatically. Because the cgroup framework don't know lifetime of cgroup_subsys_state. free_css_id() function is provided. This must be called by subsys. There are several reasons to develop this. - Saving space .... For example, memcg's swap_cgroup is array of pointers to cgroup. But it is not necessary to be very fast. By replacing pointers(8bytes per ent) to ID (2byes per ent), we can reduce much amount of memory usage. - Scanning without lock. CSS_ID provides "scan id under this ROOT" function. By this, scanning css under root can be written without locks. ex) do { rcu_read_lock(); next = cgroup_get_next(subsys, id, root, &found); /* check sanity of next here */ css_tryget(); rcu_read_unlock(); id = found + 1 } while(...) Characteristics: - Each css has unique ID under subsys. - Lifetime of ID is controlled by subsys. - css ID contains "ID" and "Depth in hierarchy" and stack of hierarchy - Allowed ID is 1-65535, ID 0 is UNUSED ID. Design Choices: - scan-by-ID v.s. scan-by-tree-walk. As /proc's pid scan does, scan-by-ID is robust when scanning is done by following kind of routine. scan -> rest a while(release a lock) -> conitunue from interrupted memcg's hierarchical reclaim does this. - When subsys->use_id is set, # of css in the system is limited to 65535. [bharata@linux.vnet.ibm.com: remove rcu_read_lock() from css_get_next()] Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Acked-by: Paul Menage <menage@google.com> Cc: Li Zefan <lizf@cn.fujitsu.com> Cc: Balbir Singh <balbir@in.ibm.com> Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp> Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-04-02 23:57:25 +00:00
idr: fix top layer handling Most functions in idr fail to deal with the high bits when the idr tree grows to the maximum height. * idr_get_empty_slot() stops growing idr tree once the depth reaches MAX_IDR_LEVEL - 1, which is one depth shallower than necessary to cover the whole range. The function doesn't even notice that it didn't grow the tree enough and ends up allocating the wrong ID given sufficiently high @starting_id. For example, on 64 bit, if the starting id is 0x7fffff01, idr_get_empty_slot() will grow the tree 5 layer deep, which only covers the 30 bits and then proceed to allocate as if the bit 30 wasn't specified. It ends up allocating 0x3fffff01 without the bit 30 but still returns 0x7fffff01. * __idr_remove_all() will not remove anything if the tree is fully grown. * idr_find() can't find anything if the tree is fully grown. * idr_for_each() and idr_get_next() can't iterate anything if the tree is fully grown. Fix it by introducing idr_max() which returns the maximum possible ID given the depth of tree and replacing the id limit checks in all affected places. As the idr_layer pointer array pa[] needs to be 1 larger than the maximum depth, enlarge pa[] arrays by one. While this plugs the discovered issues, the whole code base is horrible and in desparate need of rewrite. It's fragile like hell, Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:05:02 +00:00
while (id >= 0 && id <= max) {
cgroup: CSS ID support Patch for Per-CSS(Cgroup Subsys State) ID and private hierarchy code. This patch attaches unique ID to each css and provides following. - css_lookup(subsys, id) returns pointer to struct cgroup_subysys_state of id. - css_get_next(subsys, id, rootid, depth, foundid) returns the next css under "root" by scanning When cgroup_subsys->use_id is set, an id for css is maintained. The cgroup framework only parepares - css_id of root css for subsys - id is automatically attached at creation of css. - id is *not* freed automatically. Because the cgroup framework don't know lifetime of cgroup_subsys_state. free_css_id() function is provided. This must be called by subsys. There are several reasons to develop this. - Saving space .... For example, memcg's swap_cgroup is array of pointers to cgroup. But it is not necessary to be very fast. By replacing pointers(8bytes per ent) to ID (2byes per ent), we can reduce much amount of memory usage. - Scanning without lock. CSS_ID provides "scan id under this ROOT" function. By this, scanning css under root can be written without locks. ex) do { rcu_read_lock(); next = cgroup_get_next(subsys, id, root, &found); /* check sanity of next here */ css_tryget(); rcu_read_unlock(); id = found + 1 } while(...) Characteristics: - Each css has unique ID under subsys. - Lifetime of ID is controlled by subsys. - css ID contains "ID" and "Depth in hierarchy" and stack of hierarchy - Allowed ID is 1-65535, ID 0 is UNUSED ID. Design Choices: - scan-by-ID v.s. scan-by-tree-walk. As /proc's pid scan does, scan-by-ID is robust when scanning is done by following kind of routine. scan -> rest a while(release a lock) -> conitunue from interrupted memcg's hierarchical reclaim does this. - When subsys->use_id is set, # of css in the system is limited to 65535. [bharata@linux.vnet.ibm.com: remove rcu_read_lock() from css_get_next()] Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Acked-by: Paul Menage <menage@google.com> Cc: Li Zefan <lizf@cn.fujitsu.com> Cc: Balbir Singh <balbir@in.ibm.com> Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp> Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-04-02 23:57:25 +00:00
while (n > 0 && p) {
n -= IDR_BITS;
*paa++ = p;
p = rcu_dereference_raw(p->ary[(id >> n) & IDR_MASK]);
cgroup: CSS ID support Patch for Per-CSS(Cgroup Subsys State) ID and private hierarchy code. This patch attaches unique ID to each css and provides following. - css_lookup(subsys, id) returns pointer to struct cgroup_subysys_state of id. - css_get_next(subsys, id, rootid, depth, foundid) returns the next css under "root" by scanning When cgroup_subsys->use_id is set, an id for css is maintained. The cgroup framework only parepares - css_id of root css for subsys - id is automatically attached at creation of css. - id is *not* freed automatically. Because the cgroup framework don't know lifetime of cgroup_subsys_state. free_css_id() function is provided. This must be called by subsys. There are several reasons to develop this. - Saving space .... For example, memcg's swap_cgroup is array of pointers to cgroup. But it is not necessary to be very fast. By replacing pointers(8bytes per ent) to ID (2byes per ent), we can reduce much amount of memory usage. - Scanning without lock. CSS_ID provides "scan id under this ROOT" function. By this, scanning css under root can be written without locks. ex) do { rcu_read_lock(); next = cgroup_get_next(subsys, id, root, &found); /* check sanity of next here */ css_tryget(); rcu_read_unlock(); id = found + 1 } while(...) Characteristics: - Each css has unique ID under subsys. - Lifetime of ID is controlled by subsys. - css ID contains "ID" and "Depth in hierarchy" and stack of hierarchy - Allowed ID is 1-65535, ID 0 is UNUSED ID. Design Choices: - scan-by-ID v.s. scan-by-tree-walk. As /proc's pid scan does, scan-by-ID is robust when scanning is done by following kind of routine. scan -> rest a while(release a lock) -> conitunue from interrupted memcg's hierarchical reclaim does this. - When subsys->use_id is set, # of css in the system is limited to 65535. [bharata@linux.vnet.ibm.com: remove rcu_read_lock() from css_get_next()] Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Acked-by: Paul Menage <menage@google.com> Cc: Li Zefan <lizf@cn.fujitsu.com> Cc: Balbir Singh <balbir@in.ibm.com> Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp> Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-04-02 23:57:25 +00:00
}
if (p) {
*nextidp = id;
return p;
}
idr: fix a subtle bug in idr_get_next() The iteration logic of idr_get_next() is borrowed mostly verbatim from idr_for_each(). It walks down the tree looking for the slot matching the current ID. If the matching slot is not found, the ID is incremented by the distance of single slot at the given level and repeats. The implementation assumes that during the whole iteration id is aligned to the layer boundaries of the level closest to the leaf, which is true for all iterations starting from zero or an existing element and thus is fine for idr_for_each(). However, idr_get_next() may be given any point and if the starting id hits in the middle of a non-existent layer, increment to the next layer will end up skipping the same offset into it. For example, an IDR with IDs filled between [64, 127] would look like the following. [ 0 64 ... ] /----/ | | | NULL [ 64 ... 127 ] If idr_get_next() is called with 63 as the starting point, it will try to follow down the pointer from 0. As it is NULL, it will then try to proceed to the next slot in the same level by adding the slot distance at that level which is 64 - making the next try 127. It goes around the loop and finds and returns 127 skipping [64, 126]. Note that this bug also triggers in idr_for_each_entry() loop which deletes during iteration as deletions can make layers go away leaving the iteration with unaligned ID into missing layers. Fix it by ensuring proceeding to the next slot doesn't carry over the unaligned offset - ie. use round_up(id + 1, slot_distance) instead of id += slot_distance. Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: David Teigland <teigland@redhat.com> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:03:34 +00:00
/*
* Proceed to the next layer at the current level. Unlike
* idr_for_each(), @id isn't guaranteed to be aligned to
* layer boundary at this point and adding 1 << n may
* incorrectly skip IDs. Make sure we jump to the
* beginning of the next layer using round_up().
*/
id = round_up(id + 1, 1 << n);
cgroup: CSS ID support Patch for Per-CSS(Cgroup Subsys State) ID and private hierarchy code. This patch attaches unique ID to each css and provides following. - css_lookup(subsys, id) returns pointer to struct cgroup_subysys_state of id. - css_get_next(subsys, id, rootid, depth, foundid) returns the next css under "root" by scanning When cgroup_subsys->use_id is set, an id for css is maintained. The cgroup framework only parepares - css_id of root css for subsys - id is automatically attached at creation of css. - id is *not* freed automatically. Because the cgroup framework don't know lifetime of cgroup_subsys_state. free_css_id() function is provided. This must be called by subsys. There are several reasons to develop this. - Saving space .... For example, memcg's swap_cgroup is array of pointers to cgroup. But it is not necessary to be very fast. By replacing pointers(8bytes per ent) to ID (2byes per ent), we can reduce much amount of memory usage. - Scanning without lock. CSS_ID provides "scan id under this ROOT" function. By this, scanning css under root can be written without locks. ex) do { rcu_read_lock(); next = cgroup_get_next(subsys, id, root, &found); /* check sanity of next here */ css_tryget(); rcu_read_unlock(); id = found + 1 } while(...) Characteristics: - Each css has unique ID under subsys. - Lifetime of ID is controlled by subsys. - css ID contains "ID" and "Depth in hierarchy" and stack of hierarchy - Allowed ID is 1-65535, ID 0 is UNUSED ID. Design Choices: - scan-by-ID v.s. scan-by-tree-walk. As /proc's pid scan does, scan-by-ID is robust when scanning is done by following kind of routine. scan -> rest a while(release a lock) -> conitunue from interrupted memcg's hierarchical reclaim does this. - When subsys->use_id is set, # of css in the system is limited to 65535. [bharata@linux.vnet.ibm.com: remove rcu_read_lock() from css_get_next()] Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Acked-by: Paul Menage <menage@google.com> Cc: Li Zefan <lizf@cn.fujitsu.com> Cc: Balbir Singh <balbir@in.ibm.com> Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp> Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-04-02 23:57:25 +00:00
while (n < fls(id)) {
n += IDR_BITS;
p = *--paa;
}
}
return NULL;
}
EXPORT_SYMBOL(idr_get_next);
cgroup: CSS ID support Patch for Per-CSS(Cgroup Subsys State) ID and private hierarchy code. This patch attaches unique ID to each css and provides following. - css_lookup(subsys, id) returns pointer to struct cgroup_subysys_state of id. - css_get_next(subsys, id, rootid, depth, foundid) returns the next css under "root" by scanning When cgroup_subsys->use_id is set, an id for css is maintained. The cgroup framework only parepares - css_id of root css for subsys - id is automatically attached at creation of css. - id is *not* freed automatically. Because the cgroup framework don't know lifetime of cgroup_subsys_state. free_css_id() function is provided. This must be called by subsys. There are several reasons to develop this. - Saving space .... For example, memcg's swap_cgroup is array of pointers to cgroup. But it is not necessary to be very fast. By replacing pointers(8bytes per ent) to ID (2byes per ent), we can reduce much amount of memory usage. - Scanning without lock. CSS_ID provides "scan id under this ROOT" function. By this, scanning css under root can be written without locks. ex) do { rcu_read_lock(); next = cgroup_get_next(subsys, id, root, &found); /* check sanity of next here */ css_tryget(); rcu_read_unlock(); id = found + 1 } while(...) Characteristics: - Each css has unique ID under subsys. - Lifetime of ID is controlled by subsys. - css ID contains "ID" and "Depth in hierarchy" and stack of hierarchy - Allowed ID is 1-65535, ID 0 is UNUSED ID. Design Choices: - scan-by-ID v.s. scan-by-tree-walk. As /proc's pid scan does, scan-by-ID is robust when scanning is done by following kind of routine. scan -> rest a while(release a lock) -> conitunue from interrupted memcg's hierarchical reclaim does this. - When subsys->use_id is set, # of css in the system is limited to 65535. [bharata@linux.vnet.ibm.com: remove rcu_read_lock() from css_get_next()] Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Acked-by: Paul Menage <menage@google.com> Cc: Li Zefan <lizf@cn.fujitsu.com> Cc: Balbir Singh <balbir@in.ibm.com> Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp> Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-04-02 23:57:25 +00:00
/**
* idr_replace - replace pointer for given id
* @idp: idr handle
* @ptr: pointer you want associated with the id
* @id: lookup key
*
* Replace the pointer registered with an id and return the old value.
* A %-ENOENT return indicates that @id was not found.
* A %-EINVAL return indicates that @id was not within valid constraints.
*
* The caller must serialize with writers.
*/
void *idr_replace(struct idr *idp, void *ptr, int id)
{
int n;
struct idr_layer *p, *old_p;
if (id < 0)
idr: remove MAX_IDR_MASK and move left MAX_IDR_* into idr.c MAX_IDR_MASK is another weirdness in the idr interface. As idr covers whole positive integer range, it's defined as 0x7fffffff or INT_MAX. Its usage in idr_find(), idr_replace() and idr_remove() is bizarre. They basically mask off the sign bit and operate on the rest, so if the caller, by accident, passes in a negative number, the sign bit will be masked off and the remaining part will be used as if that was the input, which is worse than crashing. The constant is visible in idr.h and there are several users in the kernel. * drivers/i2c/i2c-core.c:i2c_add_numbered_adapter() Basically used to test if adap->nr is a negative number which isn't -1 and returns -EINVAL if so. idr_alloc() already has negative @start checking (w/ WARN_ON_ONCE), so this can go away. * drivers/infiniband/core/cm.c:cm_alloc_id() drivers/infiniband/hw/mlx4/cm.c:id_map_alloc() Used to wrap cyclic @start. Can be replaced with max(next, 0). Note that this type of cyclic allocation using idr is buggy. These are prone to spurious -ENOSPC failure after the first wraparound. * fs/super.c:get_anon_bdev() The ID allocated from ida is masked off before being tested whether it's inside valid range. ida allocated ID can never be a negative number and the masking is unnecessary. Update idr_*() functions to fail with -EINVAL when negative @id is specified and update other MAX_IDR_MASK users as described above. This leaves MAX_IDR_MASK without any user, remove it and relocate other MAX_IDR_* constants to lib/idr.c. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Jean Delvare <khali@linux-fr.org> Cc: Roland Dreier <roland@kernel.org> Cc: Sean Hefty <sean.hefty@intel.com> Cc: Hal Rosenstock <hal.rosenstock@gmail.com> Cc: "Marciniszyn, Mike" <mike.marciniszyn@intel.com> Cc: Jack Morgenstein <jackm@dev.mellanox.co.il> Cc: Or Gerlitz <ogerlitz@mellanox.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Acked-by: Wolfram Sang <wolfram@the-dreams.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:05:04 +00:00
return ERR_PTR(-EINVAL);
p = idp->top;
if (!p)
return ERR_PTR(-EINVAL);
n = (p->layer+1) * IDR_BITS;
if (id >= (1 << n))
return ERR_PTR(-EINVAL);
n -= IDR_BITS;
while ((n > 0) && p) {
p = p->ary[(id >> n) & IDR_MASK];
n -= IDR_BITS;
}
n = id & IDR_MASK;
if (unlikely(p == NULL || !test_bit(n, p->bitmap)))
return ERR_PTR(-ENOENT);
old_p = p->ary[n];
rcu_assign_pointer(p->ary[n], ptr);
return old_p;
}
EXPORT_SYMBOL(idr_replace);
void __init idr_init_cache(void)
{
idr_layer_cache = kmem_cache_create("idr_layer_cache",
lib/idr.c: use kmem_cache_zalloc() for the idr_layer cache David points out that the idr_remove_all() function returns unused slabs to the kmem cache, but needs to zero them first or else they will be uninitialized upon next use. This causes crashes which have been observed in the firewire subsystem. He fixed this by zeroing the object before freeing it in idr_remove_all(). But we agree that simply removing the constructor and zeroing the object at allocation time is simpler than relying upon slab constructor machinery and might even be faster. This problem was introduced by "idr: make idr_remove rcu-safe" (commit cf481c20c476ad2c0febdace9ce23f5a4db19582), which was first released in 2.6.27. There are no known codesites which trigger this bug in 2.6.27 or 2.6.28. The post-2.6.28 firewire changes are the only known triggerer. There might of course be not-yet-discovered triggerers in 2.6.27 and 2.6.28, and there might be out-of-tree triggerers which are added to those kernel versions. I'll let the -stable guys decide whether they want to backport this fix. Reported-by: David Moore <dcm@acm.org> Cc: Stefan Richter <stefanr@s5r6.in-berlin.de> Cc: Nadia Derbey <Nadia.Derbey@bull.net> Cc: Paul E. McKenney <paulmck@us.ibm.com> Cc: Manfred Spraul <manfred@colorfullife.com> Cc: Kristian Hgsberg <krh@redhat.com> Acked-by: Pekka Enberg <penberg@cs.helsinki.fi> Cc: <stable@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-01-15 21:51:21 +00:00
sizeof(struct idr_layer), 0, SLAB_PANIC, NULL);
}
/**
* idr_init - initialize idr handle
* @idp: idr handle
*
* This function is use to set up the handle (@idp) that you will pass
* to the rest of the functions.
*/
void idr_init(struct idr *idp)
{
memset(idp, 0, sizeof(struct idr));
spin_lock_init(&idp->lock);
}
EXPORT_SYMBOL(idr_init);
/**
* DOC: IDA description
* IDA - IDR based ID allocator
*
* This is id allocator without id -> pointer translation. Memory
* usage is much lower than full blown idr because each id only
* occupies a bit. ida uses a custom leaf node which contains
* IDA_BITMAP_BITS slots.
*
* 2007-04-25 written by Tejun Heo <htejun@gmail.com>
*/
static void free_bitmap(struct ida *ida, struct ida_bitmap *bitmap)
{
unsigned long flags;
if (!ida->free_bitmap) {
spin_lock_irqsave(&ida->idr.lock, flags);
if (!ida->free_bitmap) {
ida->free_bitmap = bitmap;
bitmap = NULL;
}
spin_unlock_irqrestore(&ida->idr.lock, flags);
}
kfree(bitmap);
}
/**
* ida_pre_get - reserve resources for ida allocation
* @ida: ida handle
* @gfp_mask: memory allocation flag
*
* This function should be called prior to locking and calling the
* following function. It preallocates enough memory to satisfy the
* worst possible allocation.
*
* If the system is REALLY out of memory this function returns %0,
* otherwise %1.
*/
int ida_pre_get(struct ida *ida, gfp_t gfp_mask)
{
/* allocate idr_layers */
if (!__idr_pre_get(&ida->idr, gfp_mask))
return 0;
/* allocate free_bitmap */
if (!ida->free_bitmap) {
struct ida_bitmap *bitmap;
bitmap = kmalloc(sizeof(struct ida_bitmap), gfp_mask);
if (!bitmap)
return 0;
free_bitmap(ida, bitmap);
}
return 1;
}
EXPORT_SYMBOL(ida_pre_get);
/**
* ida_get_new_above - allocate new ID above or equal to a start id
* @ida: ida handle
* @starting_id: id to start search at
* @p_id: pointer to the allocated handle
*
* Allocate new ID above or equal to @starting_id. It should be called
* with any required locks.
*
* If memory is required, it will return %-EAGAIN, you should unlock
* and go back to the ida_pre_get() call. If the ida is full, it will
* return %-ENOSPC.
*
* @p_id returns a value in the range @starting_id ... %0x7fffffff.
*/
int ida_get_new_above(struct ida *ida, int starting_id, int *p_id)
{
idr: fix top layer handling Most functions in idr fail to deal with the high bits when the idr tree grows to the maximum height. * idr_get_empty_slot() stops growing idr tree once the depth reaches MAX_IDR_LEVEL - 1, which is one depth shallower than necessary to cover the whole range. The function doesn't even notice that it didn't grow the tree enough and ends up allocating the wrong ID given sufficiently high @starting_id. For example, on 64 bit, if the starting id is 0x7fffff01, idr_get_empty_slot() will grow the tree 5 layer deep, which only covers the 30 bits and then proceed to allocate as if the bit 30 wasn't specified. It ends up allocating 0x3fffff01 without the bit 30 but still returns 0x7fffff01. * __idr_remove_all() will not remove anything if the tree is fully grown. * idr_find() can't find anything if the tree is fully grown. * idr_for_each() and idr_get_next() can't iterate anything if the tree is fully grown. Fix it by introducing idr_max() which returns the maximum possible ID given the depth of tree and replacing the id limit checks in all affected places. As the idr_layer pointer array pa[] needs to be 1 larger than the maximum depth, enlarge pa[] arrays by one. While this plugs the discovered issues, the whole code base is horrible and in desparate need of rewrite. It's fragile like hell, Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:05:02 +00:00
struct idr_layer *pa[MAX_IDR_LEVEL + 1];
struct ida_bitmap *bitmap;
unsigned long flags;
int idr_id = starting_id / IDA_BITMAP_BITS;
int offset = starting_id % IDA_BITMAP_BITS;
int t, id;
restart:
/* get vacant slot */
idr: implement idr_preload[_end]() and idr_alloc() The current idr interface is very cumbersome. * For all allocations, two function calls - idr_pre_get() and idr_get_new*() - should be made. * idr_pre_get() doesn't guarantee that the following idr_get_new*() will not fail from memory shortage. If idr_get_new*() returns -EAGAIN, the caller is expected to retry pre_get and allocation. * idr_get_new*() can't enforce upper limit. Upper limit can only be enforced by allocating and then freeing if above limit. * idr_layer buffer is unnecessarily per-idr. Each idr ends up keeping around MAX_IDR_FREE idr_layers. The memory consumed per idr is under two pages but it makes it difficult to make idr_layer larger. This patch implements the following new set of allocation functions. * idr_preload[_end]() - Similar to radix preload but doesn't fail. The first idr_alloc() inside preload section can be treated as if it were called with @gfp_mask used for idr_preload(). * idr_alloc() - Allocate an ID w/ lower and upper limits. Takes @gfp_flags and can be used w/o preloading. When used inside preloaded section, the allocation mask of preloading can be assumed. If idr_alloc() can be called from a context which allows sufficiently relaxed @gfp_mask, it can be used by itself. If, for example, idr_alloc() is called inside spinlock protected region, preloading can be used like the following. idr_preload(GFP_KERNEL); spin_lock(lock); id = idr_alloc(idr, ptr, start, end, GFP_NOWAIT); spin_unlock(lock); idr_preload_end(); if (id < 0) error; which is much simpler and less error-prone than idr_pre_get and idr_get_new*() loop. The new interface uses per-pcu idr_layer buffer and thus the number of idr's in the system doesn't affect the amount of memory used for preloading. idr_layer_alloc() is introduced to handle idr_layer allocations for both old and new ID allocation paths. This is a bit hairy now but the new interface is expected to replace the old and the internal implementation eventually will become simpler. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:03:55 +00:00
t = idr_get_empty_slot(&ida->idr, idr_id, pa, 0, &ida->idr);
if (t < 0)
return t == -ENOMEM ? -EAGAIN : t;
if (t * IDA_BITMAP_BITS >= MAX_IDR_BIT)
return -ENOSPC;
if (t != idr_id)
offset = 0;
idr_id = t;
/* if bitmap isn't there, create a new one */
bitmap = (void *)pa[0]->ary[idr_id & IDR_MASK];
if (!bitmap) {
spin_lock_irqsave(&ida->idr.lock, flags);
bitmap = ida->free_bitmap;
ida->free_bitmap = NULL;
spin_unlock_irqrestore(&ida->idr.lock, flags);
if (!bitmap)
return -EAGAIN;
memset(bitmap, 0, sizeof(struct ida_bitmap));
rcu_assign_pointer(pa[0]->ary[idr_id & IDR_MASK],
(void *)bitmap);
pa[0]->count++;
}
/* lookup for empty slot */
t = find_next_zero_bit(bitmap->bitmap, IDA_BITMAP_BITS, offset);
if (t == IDA_BITMAP_BITS) {
/* no empty slot after offset, continue to the next chunk */
idr_id++;
offset = 0;
goto restart;
}
id = idr_id * IDA_BITMAP_BITS + t;
if (id >= MAX_IDR_BIT)
return -ENOSPC;
__set_bit(t, bitmap->bitmap);
if (++bitmap->nr_busy == IDA_BITMAP_BITS)
idr_mark_full(pa, idr_id);
*p_id = id;
/* Each leaf node can handle nearly a thousand slots and the
* whole idea of ida is to have small memory foot print.
* Throw away extra resources one by one after each successful
* allocation.
*/
if (ida->idr.id_free_cnt || ida->free_bitmap) {
struct idr_layer *p = get_from_free_list(&ida->idr);
if (p)
kmem_cache_free(idr_layer_cache, p);
}
return 0;
}
EXPORT_SYMBOL(ida_get_new_above);
/**
* ida_remove - remove the given ID
* @ida: ida handle
* @id: ID to free
*/
void ida_remove(struct ida *ida, int id)
{
struct idr_layer *p = ida->idr.top;
int shift = (ida->idr.layers - 1) * IDR_BITS;
int idr_id = id / IDA_BITMAP_BITS;
int offset = id % IDA_BITMAP_BITS;
int n;
struct ida_bitmap *bitmap;
/* clear full bits while looking up the leaf idr_layer */
while ((shift > 0) && p) {
n = (idr_id >> shift) & IDR_MASK;
__clear_bit(n, p->bitmap);
p = p->ary[n];
shift -= IDR_BITS;
}
if (p == NULL)
goto err;
n = idr_id & IDR_MASK;
__clear_bit(n, p->bitmap);
bitmap = (void *)p->ary[n];
if (!test_bit(offset, bitmap->bitmap))
goto err;
/* update bitmap and remove it if empty */
__clear_bit(offset, bitmap->bitmap);
if (--bitmap->nr_busy == 0) {
__set_bit(n, p->bitmap); /* to please idr_remove() */
idr_remove(&ida->idr, idr_id);
free_bitmap(ida, bitmap);
}
return;
err:
printk(KERN_WARNING
"ida_remove called for id=%d which is not allocated.\n", id);
}
EXPORT_SYMBOL(ida_remove);
/**
* ida_destroy - release all cached layers within an ida tree
* @ida: ida handle
*/
void ida_destroy(struct ida *ida)
{
idr_destroy(&ida->idr);
kfree(ida->free_bitmap);
}
EXPORT_SYMBOL(ida_destroy);
/**
* ida_simple_get - get a new id.
* @ida: the (initialized) ida.
* @start: the minimum id (inclusive, < 0x8000000)
* @end: the maximum id (exclusive, < 0x8000000 or 0)
* @gfp_mask: memory allocation flags
*
* Allocates an id in the range start <= id < end, or returns -ENOSPC.
* On memory allocation failure, returns -ENOMEM.
*
* Use ida_simple_remove() to get rid of an id.
*/
int ida_simple_get(struct ida *ida, unsigned int start, unsigned int end,
gfp_t gfp_mask)
{
int ret, id;
unsigned int max;
unsigned long flags;
BUG_ON((int)start < 0);
BUG_ON((int)end < 0);
if (end == 0)
max = 0x80000000;
else {
BUG_ON(end < start);
max = end - 1;
}
again:
if (!ida_pre_get(ida, gfp_mask))
return -ENOMEM;
spin_lock_irqsave(&simple_ida_lock, flags);
ret = ida_get_new_above(ida, start, &id);
if (!ret) {
if (id > max) {
ida_remove(ida, id);
ret = -ENOSPC;
} else {
ret = id;
}
}
spin_unlock_irqrestore(&simple_ida_lock, flags);
if (unlikely(ret == -EAGAIN))
goto again;
return ret;
}
EXPORT_SYMBOL(ida_simple_get);
/**
* ida_simple_remove - remove an allocated id.
* @ida: the (initialized) ida.
* @id: the id returned by ida_simple_get.
*/
void ida_simple_remove(struct ida *ida, unsigned int id)
{
unsigned long flags;
BUG_ON((int)id < 0);
spin_lock_irqsave(&simple_ida_lock, flags);
ida_remove(ida, id);
spin_unlock_irqrestore(&simple_ida_lock, flags);
}
EXPORT_SYMBOL(ida_simple_remove);
/**
* ida_init - initialize ida handle
* @ida: ida handle
*
* This function is use to set up the handle (@ida) that you will pass
* to the rest of the functions.
*/
void ida_init(struct ida *ida)
{
memset(ida, 0, sizeof(struct ida));
idr_init(&ida->idr);
}
EXPORT_SYMBOL(ida_init);