percpu: allow pcpu_alloc() to be called with IRQs off

pcpu_alloc() and pcpu_extend_area_map() perform a series of
spin_lock_irq()/spin_unlock_irq() calls, which make them unsafe
with respect to being called from contexts which have IRQs off.

This patch converts the code to perform save/restore of flags instead,
making pcpu_alloc() (or __alloc_percpu() respectively) to be called
from early kernel startup stage, where IRQs are off.

This is needed for proper initialization of per-cpu rq_weight data from
sched_init().

tj: added comment explaining why irqsave/restore is used in alloc path.

Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Tejun Heo <tj@kernel.org>
This commit is contained in:
Jiri Kosina 2009-10-29 00:25:59 +09:00 committed by Tejun Heo
parent 1a0c3298d6
commit 403a91b165

View file

@ -153,7 +153,10 @@ static int pcpu_reserved_chunk_limit;
* *
* During allocation, pcpu_alloc_mutex is kept locked all the time and * During allocation, pcpu_alloc_mutex is kept locked all the time and
* pcpu_lock is grabbed and released as necessary. All actual memory * pcpu_lock is grabbed and released as necessary. All actual memory
* allocations are done using GFP_KERNEL with pcpu_lock released. * allocations are done using GFP_KERNEL with pcpu_lock released. In
* general, percpu memory can't be allocated with irq off but
* irqsave/restore are still used in alloc path so that it can be used
* from early init path - sched_init() specifically.
* *
* Free path accesses and alters only the index data structures, so it * Free path accesses and alters only the index data structures, so it
* can be safely called from atomic context. When memory needs to be * can be safely called from atomic context. When memory needs to be
@ -366,7 +369,7 @@ static struct pcpu_chunk *pcpu_chunk_addr_search(void *addr)
* RETURNS: * RETURNS:
* 0 if noop, 1 if successfully extended, -errno on failure. * 0 if noop, 1 if successfully extended, -errno on failure.
*/ */
static int pcpu_extend_area_map(struct pcpu_chunk *chunk) static int pcpu_extend_area_map(struct pcpu_chunk *chunk, unsigned long *flags)
{ {
int new_alloc; int new_alloc;
int *new; int *new;
@ -376,7 +379,7 @@ static int pcpu_extend_area_map(struct pcpu_chunk *chunk)
if (chunk->map_alloc >= chunk->map_used + 2) if (chunk->map_alloc >= chunk->map_used + 2)
return 0; return 0;
spin_unlock_irq(&pcpu_lock); spin_unlock_irqrestore(&pcpu_lock, *flags);
new_alloc = PCPU_DFL_MAP_ALLOC; new_alloc = PCPU_DFL_MAP_ALLOC;
while (new_alloc < chunk->map_used + 2) while (new_alloc < chunk->map_used + 2)
@ -384,7 +387,7 @@ static int pcpu_extend_area_map(struct pcpu_chunk *chunk)
new = pcpu_mem_alloc(new_alloc * sizeof(new[0])); new = pcpu_mem_alloc(new_alloc * sizeof(new[0]));
if (!new) { if (!new) {
spin_lock_irq(&pcpu_lock); spin_lock_irqsave(&pcpu_lock, *flags);
return -ENOMEM; return -ENOMEM;
} }
@ -393,7 +396,7 @@ static int pcpu_extend_area_map(struct pcpu_chunk *chunk)
* could have happened inbetween, so map_used couldn't have * could have happened inbetween, so map_used couldn't have
* grown. * grown.
*/ */
spin_lock_irq(&pcpu_lock); spin_lock_irqsave(&pcpu_lock, *flags);
BUG_ON(new_alloc < chunk->map_used + 2); BUG_ON(new_alloc < chunk->map_used + 2);
size = chunk->map_alloc * sizeof(chunk->map[0]); size = chunk->map_alloc * sizeof(chunk->map[0]);
@ -1047,6 +1050,7 @@ static void *pcpu_alloc(size_t size, size_t align, bool reserved)
struct pcpu_chunk *chunk; struct pcpu_chunk *chunk;
const char *err; const char *err;
int slot, off; int slot, off;
unsigned long flags;
if (unlikely(!size || size > PCPU_MIN_UNIT_SIZE || align > PAGE_SIZE)) { if (unlikely(!size || size > PCPU_MIN_UNIT_SIZE || align > PAGE_SIZE)) {
WARN(true, "illegal size (%zu) or align (%zu) for " WARN(true, "illegal size (%zu) or align (%zu) for "
@ -1055,13 +1059,13 @@ static void *pcpu_alloc(size_t size, size_t align, bool reserved)
} }
mutex_lock(&pcpu_alloc_mutex); mutex_lock(&pcpu_alloc_mutex);
spin_lock_irq(&pcpu_lock); spin_lock_irqsave(&pcpu_lock, flags);
/* serve reserved allocations from the reserved chunk if available */ /* serve reserved allocations from the reserved chunk if available */
if (reserved && pcpu_reserved_chunk) { if (reserved && pcpu_reserved_chunk) {
chunk = pcpu_reserved_chunk; chunk = pcpu_reserved_chunk;
if (size > chunk->contig_hint || if (size > chunk->contig_hint ||
pcpu_extend_area_map(chunk) < 0) { pcpu_extend_area_map(chunk, &flags) < 0) {
err = "failed to extend area map of reserved chunk"; err = "failed to extend area map of reserved chunk";
goto fail_unlock; goto fail_unlock;
} }
@ -1079,7 +1083,7 @@ static void *pcpu_alloc(size_t size, size_t align, bool reserved)
if (size > chunk->contig_hint) if (size > chunk->contig_hint)
continue; continue;
switch (pcpu_extend_area_map(chunk)) { switch (pcpu_extend_area_map(chunk, &flags)) {
case 0: case 0:
break; break;
case 1: case 1:
@ -1096,7 +1100,7 @@ static void *pcpu_alloc(size_t size, size_t align, bool reserved)
} }
/* hmmm... no space left, create a new chunk */ /* hmmm... no space left, create a new chunk */
spin_unlock_irq(&pcpu_lock); spin_unlock_irqrestore(&pcpu_lock, flags);
chunk = alloc_pcpu_chunk(); chunk = alloc_pcpu_chunk();
if (!chunk) { if (!chunk) {
@ -1104,16 +1108,16 @@ static void *pcpu_alloc(size_t size, size_t align, bool reserved)
goto fail_unlock_mutex; goto fail_unlock_mutex;
} }
spin_lock_irq(&pcpu_lock); spin_lock_irqsave(&pcpu_lock, flags);
pcpu_chunk_relocate(chunk, -1); pcpu_chunk_relocate(chunk, -1);
goto restart; goto restart;
area_found: area_found:
spin_unlock_irq(&pcpu_lock); spin_unlock_irqrestore(&pcpu_lock, flags);
/* populate, map and clear the area */ /* populate, map and clear the area */
if (pcpu_populate_chunk(chunk, off, size)) { if (pcpu_populate_chunk(chunk, off, size)) {
spin_lock_irq(&pcpu_lock); spin_lock_irqsave(&pcpu_lock, flags);
pcpu_free_area(chunk, off); pcpu_free_area(chunk, off);
err = "failed to populate"; err = "failed to populate";
goto fail_unlock; goto fail_unlock;
@ -1125,7 +1129,7 @@ static void *pcpu_alloc(size_t size, size_t align, bool reserved)
return __addr_to_pcpu_ptr(chunk->base_addr + off); return __addr_to_pcpu_ptr(chunk->base_addr + off);
fail_unlock: fail_unlock:
spin_unlock_irq(&pcpu_lock); spin_unlock_irqrestore(&pcpu_lock, flags);
fail_unlock_mutex: fail_unlock_mutex:
mutex_unlock(&pcpu_alloc_mutex); mutex_unlock(&pcpu_alloc_mutex);
if (warn_limit) { if (warn_limit) {