2009-10-06 22:31:10 +04:00
2010-04-07 02:14:15 +04:00
# include <linux/ceph/ceph_debug.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 11:04:11 +03:00
2010-04-07 02:14:15 +04:00
# include <linux/module.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 11:04:11 +03:00
# include <linux/slab.h>
2009-10-06 22:31:10 +04:00
# include <asm/div64.h>
2010-04-07 02:14:15 +04:00
# include <linux/ceph/libceph.h>
# include <linux/ceph/osdmap.h>
# include <linux/ceph/decode.h>
# include <linux/crush/hash.h>
# include <linux/crush/mapper.h>
2009-10-06 22:31:10 +04:00
char * ceph_osdmap_state_str ( char * str , int len , int state )
{
if ( ! len )
2013-01-26 03:48:59 +04:00
return str ;
2009-10-06 22:31:10 +04:00
2013-01-26 03:48:59 +04:00
if ( ( state & CEPH_OSD_EXISTS ) & & ( state & CEPH_OSD_UP ) )
snprintf ( str , len , " exists, up " ) ;
else if ( state & CEPH_OSD_EXISTS )
snprintf ( str , len , " exists " ) ;
else if ( state & CEPH_OSD_UP )
snprintf ( str , len , " up " ) ;
else
2009-10-06 22:31:10 +04:00
snprintf ( str , len , " doesn't exist " ) ;
2013-01-26 03:48:59 +04:00
2009-10-06 22:31:10 +04:00
return str ;
}
/* maps */
2012-04-15 09:58:06 +04:00
static int calc_bits_of ( unsigned int t )
2009-10-06 22:31:10 +04:00
{
int b = 0 ;
while ( t ) {
t = t > > 1 ;
b + + ;
}
return b ;
}
/*
* the foo_mask is the smallest value 2 ^ n - 1 that is > = foo .
*/
static void calc_pg_masks ( struct ceph_pg_pool_info * pi )
{
2013-02-23 22:41:09 +04:00
pi - > pg_num_mask = ( 1 < < calc_bits_of ( pi - > pg_num - 1 ) ) - 1 ;
pi - > pgp_num_mask = ( 1 < < calc_bits_of ( pi - > pgp_num - 1 ) ) - 1 ;
2009-10-06 22:31:10 +04:00
}
/*
* decode crush map
*/
static int crush_decode_uniform_bucket ( void * * p , void * end ,
struct crush_bucket_uniform * b )
{
dout ( " crush_decode_uniform_bucket %p to %p \n " , * p , end ) ;
ceph_decode_need ( p , end , ( 1 + b - > h . size ) * sizeof ( u32 ) , bad ) ;
2009-10-14 20:59:09 +04:00
b - > item_weight = ceph_decode_32 ( p ) ;
2009-10-06 22:31:10 +04:00
return 0 ;
bad :
return - EINVAL ;
}
static int crush_decode_list_bucket ( void * * p , void * end ,
struct crush_bucket_list * b )
{
int j ;
dout ( " crush_decode_list_bucket %p to %p \n " , * p , end ) ;
b - > item_weights = kcalloc ( b - > h . size , sizeof ( u32 ) , GFP_NOFS ) ;
if ( b - > item_weights = = NULL )
return - ENOMEM ;
b - > sum_weights = kcalloc ( b - > h . size , sizeof ( u32 ) , GFP_NOFS ) ;
if ( b - > sum_weights = = NULL )
return - ENOMEM ;
ceph_decode_need ( p , end , 2 * b - > h . size * sizeof ( u32 ) , bad ) ;
for ( j = 0 ; j < b - > h . size ; j + + ) {
2009-10-14 20:59:09 +04:00
b - > item_weights [ j ] = ceph_decode_32 ( p ) ;
b - > sum_weights [ j ] = ceph_decode_32 ( p ) ;
2009-10-06 22:31:10 +04:00
}
return 0 ;
bad :
return - EINVAL ;
}
static int crush_decode_tree_bucket ( void * * p , void * end ,
struct crush_bucket_tree * b )
{
int j ;
dout ( " crush_decode_tree_bucket %p to %p \n " , * p , end ) ;
ceph_decode_32_safe ( p , end , b - > num_nodes , bad ) ;
b - > node_weights = kcalloc ( b - > num_nodes , sizeof ( u32 ) , GFP_NOFS ) ;
if ( b - > node_weights = = NULL )
return - ENOMEM ;
ceph_decode_need ( p , end , b - > num_nodes * sizeof ( u32 ) , bad ) ;
for ( j = 0 ; j < b - > num_nodes ; j + + )
2009-10-14 20:59:09 +04:00
b - > node_weights [ j ] = ceph_decode_32 ( p ) ;
2009-10-06 22:31:10 +04:00
return 0 ;
bad :
return - EINVAL ;
}
static int crush_decode_straw_bucket ( void * * p , void * end ,
struct crush_bucket_straw * b )
{
int j ;
dout ( " crush_decode_straw_bucket %p to %p \n " , * p , end ) ;
b - > item_weights = kcalloc ( b - > h . size , sizeof ( u32 ) , GFP_NOFS ) ;
if ( b - > item_weights = = NULL )
return - ENOMEM ;
b - > straws = kcalloc ( b - > h . size , sizeof ( u32 ) , GFP_NOFS ) ;
if ( b - > straws = = NULL )
return - ENOMEM ;
ceph_decode_need ( p , end , 2 * b - > h . size * sizeof ( u32 ) , bad ) ;
for ( j = 0 ; j < b - > h . size ; j + + ) {
2009-10-14 20:59:09 +04:00
b - > item_weights [ j ] = ceph_decode_32 ( p ) ;
b - > straws [ j ] = ceph_decode_32 ( p ) ;
2009-10-06 22:31:10 +04:00
}
return 0 ;
bad :
return - EINVAL ;
}
2012-07-31 05:15:23 +04:00
static int skip_name_map ( void * * p , void * end )
{
int len ;
ceph_decode_32_safe ( p , end , len , bad ) ;
while ( len - - ) {
int strlen ;
* p + = sizeof ( u32 ) ;
ceph_decode_32_safe ( p , end , strlen , bad ) ;
* p + = strlen ;
}
return 0 ;
bad :
return - EINVAL ;
}
2009-10-06 22:31:10 +04:00
static struct crush_map * crush_decode ( void * pbyval , void * end )
{
struct crush_map * c ;
int err = - EINVAL ;
int i , j ;
void * * p = & pbyval ;
void * start = pbyval ;
u32 magic ;
2012-07-31 05:15:23 +04:00
u32 num_name_maps ;
2009-10-06 22:31:10 +04:00
dout ( " crush_decode %p to %p len %d \n " , * p , end , ( int ) ( end - * p ) ) ;
c = kzalloc ( sizeof ( * c ) , GFP_NOFS ) ;
if ( c = = NULL )
return ERR_PTR ( - ENOMEM ) ;
2012-07-31 05:15:23 +04:00
/* set tunables to default values */
c - > choose_local_tries = 2 ;
c - > choose_local_fallback_tries = 5 ;
c - > choose_total_tries = 19 ;
libceph: for chooseleaf rules, retry CRUSH map descent from root if leaf is failed
Add libceph support for a new CRUSH tunable recently added to Ceph servers.
Consider the CRUSH rule
step chooseleaf firstn 0 type <node_type>
This rule means that <n> replicas will be chosen in a manner such that
each chosen leaf's branch will contain a unique instance of <node_type>.
When an object is re-replicated after a leaf failure, if the CRUSH map uses
a chooseleaf rule the remapped replica ends up under the <node_type> bucket
that held the failed leaf. This causes uneven data distribution across the
storage cluster, to the point that when all the leaves but one fail under a
particular <node_type> bucket, that remaining leaf holds all the data from
its failed peers.
This behavior also limits the number of peers that can participate in the
re-replication of the data held by the failed leaf, which increases the
time required to re-replicate after a failure.
For a chooseleaf CRUSH rule, the tree descent has two steps: call them the
inner and outer descents.
If the tree descent down to <node_type> is the outer descent, and the descent
from <node_type> down to a leaf is the inner descent, the issue is that a
down leaf is detected on the inner descent, so only the inner descent is
retried.
In order to disperse re-replicated data as widely as possible across a
storage cluster after a failure, we want to retry the outer descent. So,
fix up crush_choose() to allow the inner descent to return immediately on
choosing a failed leaf. Wire this up as a new CRUSH tunable.
Note that after this change, for a chooseleaf rule, if the primary OSD
in a placement group has failed, choosing a replacement may result in
one of the other OSDs in the PG colliding with the new primary. This
requires that OSD's data for that PG to need moving as well. This
seems unavoidable but should be relatively rare.
This corresponds to ceph.git commit 88f218181a9e6d2292e2697fc93797d0f6d6e5dc.
Signed-off-by: Jim Schutt <jaschut@sandia.gov>
Reviewed-by: Sage Weil <sage@inktank.com>
2012-11-30 20:15:25 +04:00
c - > chooseleaf_descend_once = 0 ;
2012-07-31 05:15:23 +04:00
2009-10-06 22:31:10 +04:00
ceph_decode_need ( p , end , 4 * sizeof ( u32 ) , bad ) ;
2009-10-14 20:59:09 +04:00
magic = ceph_decode_32 ( p ) ;
2009-10-06 22:31:10 +04:00
if ( magic ! = CRUSH_MAGIC ) {
pr_err ( " crush_decode magic %x != current %x \n " ,
2012-04-15 09:58:06 +04:00
( unsigned int ) magic , ( unsigned int ) CRUSH_MAGIC ) ;
2009-10-06 22:31:10 +04:00
goto bad ;
}
2009-10-14 20:59:09 +04:00
c - > max_buckets = ceph_decode_32 ( p ) ;
c - > max_rules = ceph_decode_32 ( p ) ;
c - > max_devices = ceph_decode_32 ( p ) ;
2009-10-06 22:31:10 +04:00
c - > buckets = kcalloc ( c - > max_buckets , sizeof ( * c - > buckets ) , GFP_NOFS ) ;
if ( c - > buckets = = NULL )
goto badmem ;
c - > rules = kcalloc ( c - > max_rules , sizeof ( * c - > rules ) , GFP_NOFS ) ;
if ( c - > rules = = NULL )
goto badmem ;
/* buckets */
for ( i = 0 ; i < c - > max_buckets ; i + + ) {
int size = 0 ;
u32 alg ;
struct crush_bucket * b ;
ceph_decode_32_safe ( p , end , alg , bad ) ;
if ( alg = = 0 ) {
c - > buckets [ i ] = NULL ;
continue ;
}
dout ( " crush_decode bucket %d off %x %p to %p \n " ,
i , ( int ) ( * p - start ) , * p , end ) ;
switch ( alg ) {
case CRUSH_BUCKET_UNIFORM :
size = sizeof ( struct crush_bucket_uniform ) ;
break ;
case CRUSH_BUCKET_LIST :
size = sizeof ( struct crush_bucket_list ) ;
break ;
case CRUSH_BUCKET_TREE :
size = sizeof ( struct crush_bucket_tree ) ;
break ;
case CRUSH_BUCKET_STRAW :
size = sizeof ( struct crush_bucket_straw ) ;
break ;
default :
2009-12-22 01:49:37 +03:00
err = - EINVAL ;
2009-10-06 22:31:10 +04:00
goto bad ;
}
BUG_ON ( size = = 0 ) ;
b = c - > buckets [ i ] = kzalloc ( size , GFP_NOFS ) ;
if ( b = = NULL )
goto badmem ;
ceph_decode_need ( p , end , 4 * sizeof ( u32 ) , bad ) ;
2009-10-14 20:59:09 +04:00
b - > id = ceph_decode_32 ( p ) ;
b - > type = ceph_decode_16 ( p ) ;
2009-11-08 07:18:22 +03:00
b - > alg = ceph_decode_8 ( p ) ;
b - > hash = ceph_decode_8 ( p ) ;
2009-10-14 20:59:09 +04:00
b - > weight = ceph_decode_32 ( p ) ;
b - > size = ceph_decode_32 ( p ) ;
2009-10-06 22:31:10 +04:00
dout ( " crush_decode bucket size %d off %x %p to %p \n " ,
b - > size , ( int ) ( * p - start ) , * p , end ) ;
b - > items = kcalloc ( b - > size , sizeof ( __s32 ) , GFP_NOFS ) ;
if ( b - > items = = NULL )
goto badmem ;
b - > perm = kcalloc ( b - > size , sizeof ( u32 ) , GFP_NOFS ) ;
if ( b - > perm = = NULL )
goto badmem ;
b - > perm_n = 0 ;
ceph_decode_need ( p , end , b - > size * sizeof ( u32 ) , bad ) ;
for ( j = 0 ; j < b - > size ; j + + )
2009-10-14 20:59:09 +04:00
b - > items [ j ] = ceph_decode_32 ( p ) ;
2009-10-06 22:31:10 +04:00
switch ( b - > alg ) {
case CRUSH_BUCKET_UNIFORM :
err = crush_decode_uniform_bucket ( p , end ,
( struct crush_bucket_uniform * ) b ) ;
if ( err < 0 )
goto bad ;
break ;
case CRUSH_BUCKET_LIST :
err = crush_decode_list_bucket ( p , end ,
( struct crush_bucket_list * ) b ) ;
if ( err < 0 )
goto bad ;
break ;
case CRUSH_BUCKET_TREE :
err = crush_decode_tree_bucket ( p , end ,
( struct crush_bucket_tree * ) b ) ;
if ( err < 0 )
goto bad ;
break ;
case CRUSH_BUCKET_STRAW :
err = crush_decode_straw_bucket ( p , end ,
( struct crush_bucket_straw * ) b ) ;
if ( err < 0 )
goto bad ;
break ;
}
}
/* rules */
dout ( " rule vec is %p \n " , c - > rules ) ;
for ( i = 0 ; i < c - > max_rules ; i + + ) {
u32 yes ;
struct crush_rule * r ;
ceph_decode_32_safe ( p , end , yes , bad ) ;
if ( ! yes ) {
dout ( " crush_decode NO rule %d off %x %p to %p \n " ,
i , ( int ) ( * p - start ) , * p , end ) ;
c - > rules [ i ] = NULL ;
continue ;
}
dout ( " crush_decode rule %d off %x %p to %p \n " ,
i , ( int ) ( * p - start ) , * p , end ) ;
/* len */
ceph_decode_32_safe ( p , end , yes , bad ) ;
# if BITS_PER_LONG == 32
2009-12-22 01:49:37 +03:00
err = - EINVAL ;
2012-02-16 20:55:48 +04:00
if ( yes > ( ULONG_MAX - sizeof ( * r ) )
/ sizeof ( struct crush_rule_step ) )
2009-10-06 22:31:10 +04:00
goto bad ;
# endif
r = c - > rules [ i ] = kmalloc ( sizeof ( * r ) +
yes * sizeof ( struct crush_rule_step ) ,
GFP_NOFS ) ;
if ( r = = NULL )
goto badmem ;
dout ( " rule %d is at %p \n " , i , r ) ;
r - > len = yes ;
ceph_decode_copy_safe ( p , end , & r - > mask , 4 , bad ) ; /* 4 u8's */
ceph_decode_need ( p , end , r - > len * 3 * sizeof ( u32 ) , bad ) ;
for ( j = 0 ; j < r - > len ; j + + ) {
2009-10-14 20:59:09 +04:00
r - > steps [ j ] . op = ceph_decode_32 ( p ) ;
r - > steps [ j ] . arg1 = ceph_decode_32 ( p ) ;
r - > steps [ j ] . arg2 = ceph_decode_32 ( p ) ;
2009-10-06 22:31:10 +04:00
}
}
/* ignore trailing name maps. */
2012-07-31 05:15:23 +04:00
for ( num_name_maps = 0 ; num_name_maps < 3 ; num_name_maps + + ) {
err = skip_name_map ( p , end ) ;
if ( err < 0 )
goto done ;
}
/* tunables */
ceph_decode_need ( p , end , 3 * sizeof ( u32 ) , done ) ;
c - > choose_local_tries = ceph_decode_32 ( p ) ;
c - > choose_local_fallback_tries = ceph_decode_32 ( p ) ;
c - > choose_total_tries = ceph_decode_32 ( p ) ;
dout ( " crush decode tunable choose_local_tries = %d " ,
c - > choose_local_tries ) ;
dout ( " crush decode tunable choose_local_fallback_tries = %d " ,
c - > choose_local_fallback_tries ) ;
dout ( " crush decode tunable choose_total_tries = %d " ,
c - > choose_total_tries ) ;
2009-10-06 22:31:10 +04:00
libceph: for chooseleaf rules, retry CRUSH map descent from root if leaf is failed
Add libceph support for a new CRUSH tunable recently added to Ceph servers.
Consider the CRUSH rule
step chooseleaf firstn 0 type <node_type>
This rule means that <n> replicas will be chosen in a manner such that
each chosen leaf's branch will contain a unique instance of <node_type>.
When an object is re-replicated after a leaf failure, if the CRUSH map uses
a chooseleaf rule the remapped replica ends up under the <node_type> bucket
that held the failed leaf. This causes uneven data distribution across the
storage cluster, to the point that when all the leaves but one fail under a
particular <node_type> bucket, that remaining leaf holds all the data from
its failed peers.
This behavior also limits the number of peers that can participate in the
re-replication of the data held by the failed leaf, which increases the
time required to re-replicate after a failure.
For a chooseleaf CRUSH rule, the tree descent has two steps: call them the
inner and outer descents.
If the tree descent down to <node_type> is the outer descent, and the descent
from <node_type> down to a leaf is the inner descent, the issue is that a
down leaf is detected on the inner descent, so only the inner descent is
retried.
In order to disperse re-replicated data as widely as possible across a
storage cluster after a failure, we want to retry the outer descent. So,
fix up crush_choose() to allow the inner descent to return immediately on
choosing a failed leaf. Wire this up as a new CRUSH tunable.
Note that after this change, for a chooseleaf rule, if the primary OSD
in a placement group has failed, choosing a replacement may result in
one of the other OSDs in the PG colliding with the new primary. This
requires that OSD's data for that PG to need moving as well. This
seems unavoidable but should be relatively rare.
This corresponds to ceph.git commit 88f218181a9e6d2292e2697fc93797d0f6d6e5dc.
Signed-off-by: Jim Schutt <jaschut@sandia.gov>
Reviewed-by: Sage Weil <sage@inktank.com>
2012-11-30 20:15:25 +04:00
ceph_decode_need ( p , end , sizeof ( u32 ) , done ) ;
c - > chooseleaf_descend_once = ceph_decode_32 ( p ) ;
dout ( " crush decode tunable chooseleaf_descend_once = %d " ,
c - > chooseleaf_descend_once ) ;
2012-07-31 05:15:23 +04:00
done :
2009-10-06 22:31:10 +04:00
dout ( " crush_decode success \n " ) ;
return c ;
badmem :
err = - ENOMEM ;
bad :
dout ( " crush_decode fail %d \n " , err ) ;
crush_destroy ( c ) ;
return ERR_PTR ( err ) ;
}
/*
2010-02-17 02:53:32 +03:00
* rbtree of pg_mapping for handling pg_temp ( explicit mapping of pgid
* to a set of osds )
2009-10-06 22:31:10 +04:00
*/
2013-02-23 22:38:16 +04:00
static int pgid_cmp ( struct ceph_pg l , struct ceph_pg r )
2009-11-04 22:39:12 +03:00
{
2013-02-23 22:38:16 +04:00
if ( l . pool < r . pool )
return - 1 ;
if ( l . pool > r . pool )
return 1 ;
if ( l . seed < r . seed )
2009-11-04 22:39:12 +03:00
return - 1 ;
2013-02-23 22:38:16 +04:00
if ( l . seed > r . seed )
2009-11-04 22:39:12 +03:00
return 1 ;
return 0 ;
}
2009-10-09 09:22:37 +04:00
static int __insert_pg_mapping ( struct ceph_pg_mapping * new ,
struct rb_root * root )
2009-10-06 22:31:10 +04:00
{
struct rb_node * * p = & root - > rb_node ;
struct rb_node * parent = NULL ;
struct ceph_pg_mapping * pg = NULL ;
2009-11-04 22:39:12 +03:00
int c ;
2009-10-06 22:31:10 +04:00
2011-09-28 21:11:04 +04:00
dout ( " __insert_pg_mapping %llx %p \n " , * ( u64 * ) & new - > pgid , new ) ;
2009-10-06 22:31:10 +04:00
while ( * p ) {
parent = * p ;
pg = rb_entry ( parent , struct ceph_pg_mapping , node ) ;
2009-11-04 22:39:12 +03:00
c = pgid_cmp ( new - > pgid , pg - > pgid ) ;
if ( c < 0 )
2009-10-06 22:31:10 +04:00
p = & ( * p ) - > rb_left ;
2009-11-04 22:39:12 +03:00
else if ( c > 0 )
2009-10-06 22:31:10 +04:00
p = & ( * p ) - > rb_right ;
else
2009-10-09 09:22:37 +04:00
return - EEXIST ;
2009-10-06 22:31:10 +04:00
}
rb_link_node ( & new - > node , parent , p ) ;
rb_insert_color ( & new - > node , root ) ;
2009-10-09 09:22:37 +04:00
return 0 ;
2009-10-06 22:31:10 +04:00
}
2010-02-17 02:53:32 +03:00
static struct ceph_pg_mapping * __lookup_pg_mapping ( struct rb_root * root ,
2013-02-23 22:38:16 +04:00
struct ceph_pg pgid )
2010-02-17 02:53:32 +03:00
{
struct rb_node * n = root - > rb_node ;
struct ceph_pg_mapping * pg ;
int c ;
while ( n ) {
pg = rb_entry ( n , struct ceph_pg_mapping , node ) ;
c = pgid_cmp ( pgid , pg - > pgid ) ;
2011-09-28 21:11:04 +04:00
if ( c < 0 ) {
2010-02-17 02:53:32 +03:00
n = n - > rb_left ;
2011-09-28 21:11:04 +04:00
} else if ( c > 0 ) {
2010-02-17 02:53:32 +03:00
n = n - > rb_right ;
2011-09-28 21:11:04 +04:00
} else {
2013-02-23 22:38:16 +04:00
dout ( " __lookup_pg_mapping %lld.%x got %p \n " ,
pgid . pool , pgid . seed , pg ) ;
2010-02-17 02:53:32 +03:00
return pg ;
2011-09-28 21:11:04 +04:00
}
2010-02-17 02:53:32 +03:00
}
return NULL ;
}
2013-02-23 22:38:16 +04:00
static int __remove_pg_mapping ( struct rb_root * root , struct ceph_pg pgid )
2011-09-28 21:11:04 +04:00
{
struct ceph_pg_mapping * pg = __lookup_pg_mapping ( root , pgid ) ;
if ( pg ) {
2013-02-23 22:38:16 +04:00
dout ( " __remove_pg_mapping %lld.%x %p \n " , pgid . pool , pgid . seed ,
pg ) ;
2011-09-28 21:11:04 +04:00
rb_erase ( & pg - > node , root ) ;
kfree ( pg ) ;
return 0 ;
}
2013-02-23 22:38:16 +04:00
dout ( " __remove_pg_mapping %lld.%x dne \n " , pgid . pool , pgid . seed ) ;
2011-09-28 21:11:04 +04:00
return - ENOENT ;
}
2010-02-17 02:55:03 +03:00
/*
* rbtree of pg pool info
*/
static int __insert_pg_pool ( struct rb_root * root , struct ceph_pg_pool_info * new )
{
struct rb_node * * p = & root - > rb_node ;
struct rb_node * parent = NULL ;
struct ceph_pg_pool_info * pi = NULL ;
while ( * p ) {
parent = * p ;
pi = rb_entry ( parent , struct ceph_pg_pool_info , node ) ;
if ( new - > id < pi - > id )
p = & ( * p ) - > rb_left ;
else if ( new - > id > pi - > id )
p = & ( * p ) - > rb_right ;
else
return - EEXIST ;
}
rb_link_node ( & new - > node , parent , p ) ;
rb_insert_color ( & new - > node , root ) ;
return 0 ;
}
2013-02-23 22:41:09 +04:00
static struct ceph_pg_pool_info * __lookup_pg_pool ( struct rb_root * root , u64 id )
2010-02-17 02:55:03 +03:00
{
struct ceph_pg_pool_info * pi ;
struct rb_node * n = root - > rb_node ;
while ( n ) {
pi = rb_entry ( n , struct ceph_pg_pool_info , node ) ;
if ( id < pi - > id )
n = n - > rb_left ;
else if ( id > pi - > id )
n = n - > rb_right ;
else
return pi ;
}
return NULL ;
}
2014-01-27 19:40:19 +04:00
struct ceph_pg_pool_info * ceph_pg_pool_by_id ( struct ceph_osdmap * map , u64 id )
{
return __lookup_pg_pool ( & map - > pg_pools , id ) ;
}
2012-10-31 04:40:33 +04:00
const char * ceph_pg_pool_name_by_id ( struct ceph_osdmap * map , u64 id )
{
struct ceph_pg_pool_info * pi ;
if ( id = = CEPH_NOPOOL )
return NULL ;
if ( WARN_ON_ONCE ( id > ( u64 ) INT_MAX ) )
return NULL ;
pi = __lookup_pg_pool ( & map - > pg_pools , ( int ) id ) ;
return pi ? pi - > name : NULL ;
}
EXPORT_SYMBOL ( ceph_pg_pool_name_by_id ) ;
2010-05-17 23:31:35 +04:00
int ceph_pg_poolid_by_name ( struct ceph_osdmap * map , const char * name )
{
struct rb_node * rbp ;
for ( rbp = rb_first ( & map - > pg_pools ) ; rbp ; rbp = rb_next ( rbp ) ) {
struct ceph_pg_pool_info * pi =
rb_entry ( rbp , struct ceph_pg_pool_info , node ) ;
if ( pi - > name & & strcmp ( pi - > name , name ) = = 0 )
return pi - > id ;
}
return - ENOENT ;
}
2010-04-07 02:14:15 +04:00
EXPORT_SYMBOL ( ceph_pg_poolid_by_name ) ;
2010-05-17 23:31:35 +04:00
2010-04-10 02:46:42 +04:00
static void __remove_pg_pool ( struct rb_root * root , struct ceph_pg_pool_info * pi )
{
rb_erase ( & pi - > node , root ) ;
kfree ( pi - > name ) ;
kfree ( pi ) ;
}
2010-08-02 22:00:55 +04:00
static int __decode_pool ( void * * p , void * end , struct ceph_pg_pool_info * pi )
2010-03-17 20:05:28 +03:00
{
2013-02-23 22:41:09 +04:00
u8 ev , cv ;
unsigned len , num ;
void * pool_end ;
ceph_decode_need ( p , end , 2 + 4 , bad ) ;
ev = ceph_decode_8 ( p ) ; /* encoding version */
cv = ceph_decode_8 ( p ) ; /* compat version */
if ( ev < 5 ) {
pr_warning ( " got v %d < 5 cv %d of ceph_pg_pool \n " , ev , cv ) ;
return - EINVAL ;
}
2014-01-27 19:40:19 +04:00
if ( cv > 9 ) {
pr_warning ( " got v %d cv %d > 9 of ceph_pg_pool \n " , ev , cv ) ;
2013-02-23 22:41:09 +04:00
return - EINVAL ;
}
len = ceph_decode_32 ( p ) ;
ceph_decode_need ( p , end , len , bad ) ;
pool_end = * p + len ;
2010-08-02 22:00:55 +04:00
2013-02-23 22:41:09 +04:00
pi - > type = ceph_decode_8 ( p ) ;
pi - > size = ceph_decode_8 ( p ) ;
pi - > crush_ruleset = ceph_decode_8 ( p ) ;
pi - > object_hash = ceph_decode_8 ( p ) ;
2010-08-02 22:00:55 +04:00
2013-02-23 22:41:09 +04:00
pi - > pg_num = ceph_decode_32 ( p ) ;
pi - > pgp_num = ceph_decode_32 ( p ) ;
* p + = 4 + 4 ; /* skip lpg* */
* p + = 4 ; /* skip last_change */
* p + = 8 + 4 ; /* skip snap_seq, snap_epoch */
/* skip snaps */
num = ceph_decode_32 ( p ) ;
while ( num - - ) {
* p + = 8 ; /* snapid key */
* p + = 1 + 1 ; /* versions */
len = ceph_decode_32 ( p ) ;
* p + = len ;
2010-08-02 22:00:55 +04:00
}
2014-01-27 19:40:19 +04:00
/* skip removed_snaps */
2013-02-23 22:41:09 +04:00
num = ceph_decode_32 ( p ) ;
* p + = num * ( 8 + 8 ) ;
* p + = 8 ; /* skip auid */
pi - > flags = ceph_decode_64 ( p ) ;
2014-01-27 19:40:19 +04:00
* p + = 4 ; /* skip crash_replay_interval */
if ( ev > = 7 )
* p + = 1 ; /* skip min_size */
if ( ev > = 8 )
* p + = 8 + 8 ; /* skip quota_max_* */
if ( ev > = 9 ) {
/* skip tiers */
num = ceph_decode_32 ( p ) ;
* p + = num * 8 ;
* p + = 8 ; /* skip tier_of */
* p + = 1 ; /* skip cache_mode */
pi - > read_tier = ceph_decode_64 ( p ) ;
pi - > write_tier = ceph_decode_64 ( p ) ;
} else {
pi - > read_tier = - 1 ;
pi - > write_tier = - 1 ;
}
2013-02-23 22:41:09 +04:00
/* ignore the rest */
* p = pool_end ;
calc_pg_masks ( pi ) ;
2010-08-02 22:00:55 +04:00
return 0 ;
bad :
return - EINVAL ;
2010-03-17 20:05:28 +03:00
}
2010-04-10 02:46:42 +04:00
static int __decode_pool_names ( void * * p , void * end , struct ceph_osdmap * map )
{
struct ceph_pg_pool_info * pi ;
2013-02-23 22:41:09 +04:00
u32 num , len ;
u64 pool ;
2010-04-10 02:46:42 +04:00
ceph_decode_32_safe ( p , end , num , bad ) ;
dout ( " %d pool names \n " , num ) ;
while ( num - - ) {
2013-02-23 22:41:09 +04:00
ceph_decode_64_safe ( p , end , pool , bad ) ;
2010-04-10 02:46:42 +04:00
ceph_decode_32_safe ( p , end , len , bad ) ;
2013-02-23 22:41:09 +04:00
dout ( " pool %llu len %d \n " , pool , len ) ;
2012-06-07 04:35:55 +04:00
ceph_decode_need ( p , end , len , bad ) ;
2010-04-10 02:46:42 +04:00
pi = __lookup_pg_pool ( & map - > pg_pools , pool ) ;
if ( pi ) {
2012-06-07 04:35:55 +04:00
char * name = kstrndup ( * p , len , GFP_NOFS ) ;
if ( ! name )
return - ENOMEM ;
2010-04-10 02:46:42 +04:00
kfree ( pi - > name ) ;
2012-06-07 04:35:55 +04:00
pi - > name = name ;
dout ( " name is %s \n " , pi - > name ) ;
2010-04-10 02:46:42 +04:00
}
* p + = len ;
}
return 0 ;
bad :
return - EINVAL ;
}
/*
* osd map
*/
void ceph_osdmap_destroy ( struct ceph_osdmap * map )
{
dout ( " osdmap_destroy %p \n " , map ) ;
if ( map - > crush )
crush_destroy ( map - > crush ) ;
while ( ! RB_EMPTY_ROOT ( & map - > pg_temp ) ) {
struct ceph_pg_mapping * pg =
rb_entry ( rb_first ( & map - > pg_temp ) ,
struct ceph_pg_mapping , node ) ;
rb_erase ( & pg - > node , & map - > pg_temp ) ;
kfree ( pg ) ;
}
while ( ! RB_EMPTY_ROOT ( & map - > pg_pools ) ) {
struct ceph_pg_pool_info * pi =
rb_entry ( rb_first ( & map - > pg_pools ) ,
struct ceph_pg_pool_info , node ) ;
__remove_pg_pool ( & map - > pg_pools , pi ) ;
}
kfree ( map - > osd_state ) ;
kfree ( map - > osd_weight ) ;
kfree ( map - > osd_addr ) ;
kfree ( map ) ;
}
/*
* adjust max osd value . reallocate arrays .
*/
static int osdmap_set_max_osd ( struct ceph_osdmap * map , int max )
{
u8 * state ;
struct ceph_entity_addr * addr ;
u32 * weight ;
state = kcalloc ( max , sizeof ( * state ) , GFP_NOFS ) ;
addr = kcalloc ( max , sizeof ( * addr ) , GFP_NOFS ) ;
weight = kcalloc ( max , sizeof ( * weight ) , GFP_NOFS ) ;
if ( state = = NULL | | addr = = NULL | | weight = = NULL ) {
kfree ( state ) ;
kfree ( addr ) ;
kfree ( weight ) ;
return - ENOMEM ;
}
/* copy old? */
if ( map - > osd_state ) {
memcpy ( state , map - > osd_state , map - > max_osd * sizeof ( * state ) ) ;
memcpy ( addr , map - > osd_addr , map - > max_osd * sizeof ( * addr ) ) ;
memcpy ( weight , map - > osd_weight , map - > max_osd * sizeof ( * weight ) ) ;
kfree ( map - > osd_state ) ;
kfree ( map - > osd_addr ) ;
kfree ( map - > osd_weight ) ;
}
map - > osd_state = state ;
map - > osd_weight = weight ;
map - > osd_addr = addr ;
map - > max_osd = max ;
return 0 ;
}
2009-10-06 22:31:10 +04:00
/*
* decode a full map .
*/
2014-03-13 18:36:13 +04:00
static int osdmap_decode ( void * * p , void * end , struct ceph_osdmap * map )
2009-10-06 22:31:10 +04:00
{
u16 version ;
2014-03-13 18:36:13 +04:00
u32 epoch = 0 ;
2009-10-06 22:31:10 +04:00
void * start = * p ;
2014-03-13 18:36:14 +04:00
u32 max ;
u32 len , i ;
2014-03-13 18:36:14 +04:00
int err ;
2010-02-17 02:55:03 +03:00
struct ceph_pg_pool_info * pi ;
2009-10-06 22:31:10 +04:00
2014-03-13 18:36:13 +04:00
dout ( " %s %p to %p len %d \n " , __func__ , * p , end , ( int ) ( end - * p ) ) ;
2009-10-06 22:31:10 +04:00
2014-03-13 18:36:14 +04:00
ceph_decode_16_safe ( p , end , version , e_inval ) ;
2013-02-23 22:41:09 +04:00
if ( version > 6 ) {
pr_warning ( " got unknown v %d > 6 of osdmap \n " , version ) ;
2014-03-13 18:36:14 +04:00
goto e_inval ;
2013-02-23 22:41:09 +04:00
}
if ( version < 6 ) {
pr_warning ( " got old v %d < 6 of osdmap \n " , version ) ;
2014-03-13 18:36:14 +04:00
goto e_inval ;
2010-02-05 03:18:10 +03:00
}
2009-10-06 22:31:10 +04:00
2014-03-13 18:36:14 +04:00
ceph_decode_need ( p , end , 2 * sizeof ( u64 ) + 6 * sizeof ( u32 ) , e_inval ) ;
2009-10-06 22:31:10 +04:00
ceph_decode_copy ( p , & map - > fsid , sizeof ( map - > fsid ) ) ;
2014-03-13 18:36:13 +04:00
epoch = map - > epoch = ceph_decode_32 ( p ) ;
2009-10-06 22:31:10 +04:00
ceph_decode_copy ( p , & map - > created , sizeof ( map - > created ) ) ;
ceph_decode_copy ( p , & map - > modified , sizeof ( map - > modified ) ) ;
2014-03-13 18:36:14 +04:00
ceph_decode_32_safe ( p , end , max , e_inval ) ;
2009-10-06 22:31:10 +04:00
while ( max - - ) {
2014-03-13 18:36:14 +04:00
ceph_decode_need ( p , end , 8 + 2 , e_inval ) ;
2010-04-10 02:46:42 +04:00
pi = kzalloc ( sizeof ( * pi ) , GFP_NOFS ) ;
2014-03-13 18:36:14 +04:00
if ( ! pi ) {
err = - ENOMEM ;
2009-10-06 22:31:10 +04:00
goto bad ;
2014-03-13 18:36:14 +04:00
}
2013-02-23 22:41:09 +04:00
pi - > id = ceph_decode_64 ( p ) ;
2010-08-02 22:00:55 +04:00
err = __decode_pool ( p , end , pi ) ;
2010-12-25 01:01:12 +03:00
if ( err < 0 ) {
kfree ( pi ) ;
2010-08-02 22:00:55 +04:00
goto bad ;
2010-12-25 01:01:12 +03:00
}
2010-02-17 02:55:03 +03:00
__insert_pg_pool ( & map - > pg_pools , pi ) ;
2009-10-06 22:31:10 +04:00
}
2010-04-10 02:46:42 +04:00
2013-02-23 22:41:09 +04:00
err = __decode_pool_names ( p , end , map ) ;
2014-03-13 18:36:14 +04:00
if ( err )
2013-02-23 22:41:09 +04:00
goto bad ;
2010-04-10 02:46:42 +04:00
2014-03-13 18:36:14 +04:00
ceph_decode_32_safe ( p , end , map - > pool_max , e_inval ) ;
2009-10-06 22:31:10 +04:00
2014-03-13 18:36:14 +04:00
ceph_decode_32_safe ( p , end , map - > flags , e_inval ) ;
2009-10-06 22:31:10 +04:00
2014-03-13 18:36:14 +04:00
/* max_osd */
ceph_decode_32_safe ( p , end , max , e_inval ) ;
2009-10-06 22:31:10 +04:00
/* (re)alloc osd arrays */
err = osdmap_set_max_osd ( map , max ) ;
2014-03-13 18:36:14 +04:00
if ( err )
2009-10-06 22:31:10 +04:00
goto bad ;
/* osds */
ceph_decode_need ( p , end , 3 * sizeof ( u32 ) +
map - > max_osd * ( 1 + sizeof ( * map - > osd_weight ) +
2014-03-13 18:36:14 +04:00
sizeof ( * map - > osd_addr ) ) , e_inval ) ;
2009-10-06 22:31:10 +04:00
* p + = 4 ; /* skip length field (should match max) */
ceph_decode_copy ( p , map - > osd_state , map - > max_osd ) ;
* p + = 4 ; /* skip length field (should match max) */
for ( i = 0 ; i < map - > max_osd ; i + + )
2009-10-14 20:59:09 +04:00
map - > osd_weight [ i ] = ceph_decode_32 ( p ) ;
2009-10-06 22:31:10 +04:00
* p + = 4 ; /* skip length field (should match max) */
ceph_decode_copy ( p , map - > osd_addr , map - > max_osd * sizeof ( * map - > osd_addr ) ) ;
2009-11-04 02:17:56 +03:00
for ( i = 0 ; i < map - > max_osd ; i + + )
ceph_decode_addr ( & map - > osd_addr [ i ] ) ;
2009-10-06 22:31:10 +04:00
/* pg_temp */
2014-03-13 18:36:14 +04:00
ceph_decode_32_safe ( p , end , len , e_inval ) ;
2009-10-06 22:31:10 +04:00
for ( i = 0 ; i < len ; i + + ) {
int n , j ;
2013-02-23 22:38:16 +04:00
struct ceph_pg pgid ;
2009-10-06 22:31:10 +04:00
struct ceph_pg_mapping * pg ;
2013-04-02 03:58:26 +04:00
err = ceph_decode_pgid ( p , end , & pgid ) ;
2013-03-07 02:57:03 +04:00
if ( err )
goto bad ;
2014-03-13 18:36:14 +04:00
ceph_decode_need ( p , end , sizeof ( u32 ) , e_inval ) ;
2009-10-14 20:59:09 +04:00
n = ceph_decode_32 ( p ) ;
2012-06-07 04:35:55 +04:00
if ( n > ( UINT_MAX - sizeof ( * pg ) ) / sizeof ( u32 ) )
2014-03-13 18:36:14 +04:00
goto e_inval ;
ceph_decode_need ( p , end , n * sizeof ( u32 ) , e_inval ) ;
2009-10-06 22:31:10 +04:00
pg = kmalloc ( sizeof ( * pg ) + n * sizeof ( u32 ) , GFP_NOFS ) ;
2014-03-13 18:36:14 +04:00
if ( ! pg ) {
err = - ENOMEM ;
2009-10-06 22:31:10 +04:00
goto bad ;
2014-03-13 18:36:14 +04:00
}
2009-10-06 22:31:10 +04:00
pg - > pgid = pgid ;
pg - > len = n ;
for ( j = 0 ; j < n ; j + + )
2009-10-14 20:59:09 +04:00
pg - > osds [ j ] = ceph_decode_32 ( p ) ;
2009-10-06 22:31:10 +04:00
2009-10-09 09:22:37 +04:00
err = __insert_pg_mapping ( pg , & map - > pg_temp ) ;
if ( err )
goto bad ;
2013-02-23 22:38:16 +04:00
dout ( " added pg_temp %lld.%x len %d \n " , pgid . pool , pgid . seed ,
len ) ;
2009-10-06 22:31:10 +04:00
}
/* crush */
2014-03-13 18:36:14 +04:00
ceph_decode_32_safe ( p , end , len , e_inval ) ;
2009-10-06 22:31:10 +04:00
dout ( " osdmap_decode crush len %d from off 0x%x \n " , len ,
( int ) ( * p - start ) ) ;
2014-03-13 18:36:14 +04:00
ceph_decode_need ( p , end , len , e_inval ) ;
2009-10-06 22:31:10 +04:00
map - > crush = crush_decode ( * p , end ) ;
* p + = len ;
if ( IS_ERR ( map - > crush ) ) {
err = PTR_ERR ( map - > crush ) ;
map - > crush = NULL ;
goto bad ;
}
2014-03-13 18:36:13 +04:00
/* ignore the rest */
2009-10-06 22:31:10 +04:00
* p = end ;
2014-03-13 18:36:13 +04:00
dout ( " full osdmap epoch %d max_osd %d \n " , map - > epoch , map - > max_osd ) ;
2014-03-13 18:36:13 +04:00
return 0 ;
2009-10-06 22:31:10 +04:00
2014-03-13 18:36:14 +04:00
e_inval :
err = - EINVAL ;
2009-10-06 22:31:10 +04:00
bad :
2014-03-13 18:36:13 +04:00
pr_err ( " corrupt full osdmap (%d) epoch %d off %d (%p of %p-%p) \n " ,
err , epoch , ( int ) ( * p - start ) , * p , start , end ) ;
print_hex_dump ( KERN_DEBUG , " osdmap: " ,
DUMP_PREFIX_OFFSET , 16 , 1 ,
start , end - start , true ) ;
2014-03-13 18:36:13 +04:00
return err ;
}
/*
* Allocate and decode a full map .
*/
struct ceph_osdmap * ceph_osdmap_decode ( void * * p , void * end )
{
struct ceph_osdmap * map ;
int ret ;
map = kzalloc ( sizeof ( * map ) , GFP_NOFS ) ;
if ( ! map )
return ERR_PTR ( - ENOMEM ) ;
map - > pg_temp = RB_ROOT ;
mutex_init ( & map - > crush_scratch_mutex ) ;
ret = osdmap_decode ( p , end , map ) ;
if ( ret ) {
ceph_osdmap_destroy ( map ) ;
return ERR_PTR ( ret ) ;
}
return map ;
2009-10-06 22:31:10 +04:00
}
/*
* decode and apply an incremental map update .
*/
struct ceph_osdmap * osdmap_apply_incremental ( void * * p , void * end ,
struct ceph_osdmap * map ,
struct ceph_messenger * msgr )
{
struct crush_map * newcrush = NULL ;
struct ceph_fsid fsid ;
u32 epoch = 0 ;
struct ceph_timespec modified ;
2013-02-23 22:41:09 +04:00
s32 len ;
u64 pool ;
__s64 new_pool_max ;
__s32 new_flags , max ;
2009-10-06 22:31:10 +04:00
void * start = * p ;
int err = - EINVAL ;
u16 version ;
2014-03-13 18:36:13 +04:00
dout ( " %s %p to %p len %d \n " , __func__ , * p , end , ( int ) ( end - * p ) ) ;
2009-10-06 22:31:10 +04:00
ceph_decode_16_safe ( p , end , version , bad ) ;
2013-03-07 02:57:03 +04:00
if ( version ! = 6 ) {
pr_warning ( " got unknown v %d != 6 of inc osdmap \n " , version ) ;
2010-02-05 03:18:10 +03:00
goto bad ;
}
2009-10-06 22:31:10 +04:00
ceph_decode_need ( p , end , sizeof ( fsid ) + sizeof ( modified ) + 2 * sizeof ( u32 ) ,
bad ) ;
ceph_decode_copy ( p , & fsid , sizeof ( fsid ) ) ;
2009-10-14 20:59:09 +04:00
epoch = ceph_decode_32 ( p ) ;
2009-10-06 22:31:10 +04:00
BUG_ON ( epoch ! = map - > epoch + 1 ) ;
ceph_decode_copy ( p , & modified , sizeof ( modified ) ) ;
2013-02-23 22:41:09 +04:00
new_pool_max = ceph_decode_64 ( p ) ;
2009-10-14 20:59:09 +04:00
new_flags = ceph_decode_32 ( p ) ;
2009-10-06 22:31:10 +04:00
/* full map? */
ceph_decode_32_safe ( p , end , len , bad ) ;
if ( len > 0 ) {
dout ( " apply_incremental full map len %d, %p to %p \n " ,
len , * p , end ) ;
2014-03-13 18:36:13 +04:00
return ceph_osdmap_decode ( p , min ( * p + len , end ) ) ;
2009-10-06 22:31:10 +04:00
}
/* new crush? */
ceph_decode_32_safe ( p , end , len , bad ) ;
if ( len > 0 ) {
dout ( " apply_incremental new crush map len %d, %p to %p \n " ,
len , * p , end ) ;
newcrush = crush_decode ( * p , min ( * p + len , end ) ) ;
if ( IS_ERR ( newcrush ) )
2010-05-22 14:01:14 +04:00
return ERR_CAST ( newcrush ) ;
2010-06-17 21:22:48 +04:00
* p + = len ;
2009-10-06 22:31:10 +04:00
}
/* new flags? */
if ( new_flags > = 0 )
map - > flags = new_flags ;
2010-02-17 02:55:03 +03:00
if ( new_pool_max > = 0 )
map - > pool_max = new_pool_max ;
2009-10-06 22:31:10 +04:00
ceph_decode_need ( p , end , 5 * sizeof ( u32 ) , bad ) ;
/* new max? */
2009-10-14 20:59:09 +04:00
max = ceph_decode_32 ( p ) ;
2009-10-06 22:31:10 +04:00
if ( max > = 0 ) {
err = osdmap_set_max_osd ( map , max ) ;
if ( err < 0 )
goto bad ;
}
map - > epoch + + ;
2011-05-13 02:18:43 +04:00
map - > modified = modified ;
2009-10-06 22:31:10 +04:00
if ( newcrush ) {
if ( map - > crush )
crush_destroy ( map - > crush ) ;
map - > crush = newcrush ;
newcrush = NULL ;
}
/* new_pool */
ceph_decode_32_safe ( p , end , len , bad ) ;
while ( len - - ) {
2010-02-17 02:55:03 +03:00
struct ceph_pg_pool_info * pi ;
2010-01-26 03:03:02 +03:00
2013-02-23 22:41:09 +04:00
ceph_decode_64_safe ( p , end , pool , bad ) ;
2010-02-17 02:55:03 +03:00
pi = __lookup_pg_pool ( & map - > pg_pools , pool ) ;
if ( ! pi ) {
2010-04-10 02:46:42 +04:00
pi = kzalloc ( sizeof ( * pi ) , GFP_NOFS ) ;
2010-02-17 02:55:03 +03:00
if ( ! pi ) {
err = - ENOMEM ;
goto bad ;
}
pi - > id = pool ;
__insert_pg_pool ( & map - > pg_pools , pi ) ;
}
2010-08-02 22:00:55 +04:00
err = __decode_pool ( p , end , pi ) ;
if ( err < 0 )
goto bad ;
2009-10-06 22:31:10 +04:00
}
2012-10-29 22:01:42 +04:00
if ( version > = 5 ) {
err = __decode_pool_names ( p , end , map ) ;
if ( err < 0 )
goto bad ;
}
2009-10-06 22:31:10 +04:00
2010-02-17 02:55:03 +03:00
/* old_pool */
2009-10-06 22:31:10 +04:00
ceph_decode_32_safe ( p , end , len , bad ) ;
2010-02-17 02:55:03 +03:00
while ( len - - ) {
struct ceph_pg_pool_info * pi ;
2013-02-23 22:41:09 +04:00
ceph_decode_64_safe ( p , end , pool , bad ) ;
2010-02-17 02:55:03 +03:00
pi = __lookup_pg_pool ( & map - > pg_pools , pool ) ;
2010-04-10 02:46:42 +04:00
if ( pi )
__remove_pg_pool ( & map - > pg_pools , pi ) ;
2010-02-17 02:55:03 +03:00
}
2009-10-06 22:31:10 +04:00
/* new_up */
err = - EINVAL ;
ceph_decode_32_safe ( p , end , len , bad ) ;
while ( len - - ) {
u32 osd ;
struct ceph_entity_addr addr ;
ceph_decode_32_safe ( p , end , osd , bad ) ;
ceph_decode_copy_safe ( p , end , & addr , sizeof ( addr ) , bad ) ;
2009-11-04 02:17:56 +03:00
ceph_decode_addr ( & addr ) ;
2009-10-06 22:31:10 +04:00
pr_info ( " osd%d up \n " , osd ) ;
BUG_ON ( osd > = map - > max_osd ) ;
map - > osd_state [ osd ] | = CEPH_OSD_UP ;
map - > osd_addr [ osd ] = addr ;
}
2011-05-03 23:52:05 +04:00
/* new_state */
2009-10-06 22:31:10 +04:00
ceph_decode_32_safe ( p , end , len , bad ) ;
while ( len - - ) {
u32 osd ;
2011-05-03 23:52:05 +04:00
u8 xorstate ;
2009-10-06 22:31:10 +04:00
ceph_decode_32_safe ( p , end , osd , bad ) ;
2011-05-03 23:52:05 +04:00
xorstate = * * ( u8 * * ) p ;
2009-10-06 22:31:10 +04:00
( * p ) + + ; /* clean flag */
2011-05-03 23:52:05 +04:00
if ( xorstate = = 0 )
xorstate = CEPH_OSD_UP ;
if ( xorstate & CEPH_OSD_UP )
pr_info ( " osd%d down \n " , osd ) ;
2009-10-06 22:31:10 +04:00
if ( osd < map - > max_osd )
2011-05-03 23:52:05 +04:00
map - > osd_state [ osd ] ^ = xorstate ;
2009-10-06 22:31:10 +04:00
}
/* new_weight */
ceph_decode_32_safe ( p , end , len , bad ) ;
while ( len - - ) {
u32 osd , off ;
ceph_decode_need ( p , end , sizeof ( u32 ) * 2 , bad ) ;
2009-10-14 20:59:09 +04:00
osd = ceph_decode_32 ( p ) ;
off = ceph_decode_32 ( p ) ;
2009-10-06 22:31:10 +04:00
pr_info ( " osd%d weight 0x%x %s \n " , osd , off ,
off = = CEPH_OSD_IN ? " (in) " :
( off = = CEPH_OSD_OUT ? " (out) " : " " ) ) ;
if ( osd < map - > max_osd )
map - > osd_weight [ osd ] = off ;
}
/* new_pg_temp */
ceph_decode_32_safe ( p , end , len , bad ) ;
while ( len - - ) {
struct ceph_pg_mapping * pg ;
int j ;
2013-02-23 22:38:16 +04:00
struct ceph_pg pgid ;
2009-10-06 22:31:10 +04:00
u32 pglen ;
2013-04-02 03:58:26 +04:00
err = ceph_decode_pgid ( p , end , & pgid ) ;
2013-03-07 02:57:03 +04:00
if ( err )
goto bad ;
ceph_decode_need ( p , end , sizeof ( u32 ) , bad ) ;
pglen = ceph_decode_32 ( p ) ;
2009-10-06 22:31:10 +04:00
if ( pglen ) {
ceph_decode_need ( p , end , pglen * sizeof ( u32 ) , bad ) ;
2012-05-21 20:45:23 +04:00
/* removing existing (if any) */
( void ) __remove_pg_mapping ( & map - > pg_temp , pgid ) ;
/* insert */
2012-10-29 22:01:42 +04:00
err = - EINVAL ;
if ( pglen > ( UINT_MAX - sizeof ( * pg ) ) / sizeof ( u32 ) )
2012-06-07 04:35:55 +04:00
goto bad ;
2012-10-29 22:01:42 +04:00
err = - ENOMEM ;
2009-10-06 22:31:10 +04:00
pg = kmalloc ( sizeof ( * pg ) + sizeof ( u32 ) * pglen , GFP_NOFS ) ;
2012-10-29 22:01:42 +04:00
if ( ! pg )
2009-10-06 22:31:10 +04:00
goto bad ;
pg - > pgid = pgid ;
pg - > len = pglen ;
2009-12-22 03:02:37 +03:00
for ( j = 0 ; j < pglen ; j + + )
2009-10-14 20:59:09 +04:00
pg - > osds [ j ] = ceph_decode_32 ( p ) ;
2009-10-09 09:22:37 +04:00
err = __insert_pg_mapping ( pg , & map - > pg_temp ) ;
2010-07-21 03:19:56 +04:00
if ( err ) {
kfree ( pg ) ;
2009-10-09 09:22:37 +04:00
goto bad ;
2010-07-21 03:19:56 +04:00
}
2013-02-23 22:38:16 +04:00
dout ( " added pg_temp %lld.%x len %d \n " , pgid . pool ,
pgid . seed , pglen ) ;
2011-09-28 21:11:04 +04:00
} else {
/* remove */
__remove_pg_mapping ( & map - > pg_temp , pgid ) ;
2009-10-06 22:31:10 +04:00
}
}
/* ignore the rest */
* p = end ;
2014-03-13 18:36:13 +04:00
dout ( " inc osdmap epoch %d max_osd %d \n " , map - > epoch , map - > max_osd ) ;
2009-10-06 22:31:10 +04:00
return map ;
bad :
2014-03-13 18:36:13 +04:00
pr_err ( " corrupt inc osdmap (%d) epoch %d off %d (%p of %p-%p) \n " ,
err , epoch , ( int ) ( * p - start ) , * p , start , end ) ;
2009-12-15 02:13:47 +03:00
print_hex_dump ( KERN_DEBUG , " osdmap: " ,
DUMP_PREFIX_OFFSET , 16 , 1 ,
start , end - start , true ) ;
2009-10-06 22:31:10 +04:00
if ( newcrush )
crush_destroy ( newcrush ) ;
return ERR_PTR ( err ) ;
}
/*
* calculate file layout from given offset , length .
* fill in correct oid , logical length , and object extent
* offset , length .
*
* for now , we write only a single su , until we can
* pass a stride back to the caller .
*/
2012-09-25 07:59:48 +04:00
int ceph_calc_file_object_mapping ( struct ceph_file_layout * layout ,
2012-11-14 19:38:19 +04:00
u64 off , u64 len ,
2009-10-29 01:15:05 +03:00
u64 * ono ,
2009-10-06 22:31:10 +04:00
u64 * oxoff , u64 * oxlen )
{
u32 osize = le32_to_cpu ( layout - > fl_object_size ) ;
u32 su = le32_to_cpu ( layout - > fl_stripe_unit ) ;
u32 sc = le32_to_cpu ( layout - > fl_stripe_count ) ;
u32 bl , stripeno , stripepos , objsetno ;
u32 su_per_object ;
2009-10-30 22:57:30 +03:00
u64 t , su_offset ;
2009-10-06 22:31:10 +04:00
2012-11-14 19:38:19 +04:00
dout ( " mapping %llu~%llu osize %u fl_su %u \n " , off , len ,
2009-10-06 22:31:10 +04:00
osize , su ) ;
2012-09-25 07:59:48 +04:00
if ( su = = 0 | | sc = = 0 )
goto invalid ;
2009-10-29 00:04:48 +03:00
su_per_object = osize / su ;
2012-09-25 07:59:48 +04:00
if ( su_per_object = = 0 )
goto invalid ;
2009-10-06 22:31:10 +04:00
dout ( " osize %u / su %u = su_per_object %u \n " , osize , su ,
su_per_object ) ;
2012-09-25 07:59:48 +04:00
if ( ( su & ~ PAGE_MASK ) ! = 0 )
goto invalid ;
2009-10-06 22:31:10 +04:00
/* bl = *off / su; */
t = off ;
do_div ( t , su ) ;
bl = t ;
dout ( " off %llu / su %u = bl %u \n " , off , su , bl ) ;
stripeno = bl / sc ;
stripepos = bl % sc ;
objsetno = stripeno / su_per_object ;
2009-10-29 01:15:05 +03:00
* ono = objsetno * sc + stripepos ;
2012-04-15 09:58:06 +04:00
dout ( " objset %u * sc %u = ono %u \n " , objsetno , sc , ( unsigned int ) * ono ) ;
2009-10-29 01:15:05 +03:00
/* *oxoff = *off % layout->fl_stripe_unit; # offset in su */
2009-10-06 22:31:10 +04:00
t = off ;
2009-10-30 22:57:30 +03:00
su_offset = do_div ( t , su ) ;
* oxoff = su_offset + ( stripeno % su_per_object ) * su ;
/*
* Calculate the length of the extent being written to the selected
2012-11-14 19:38:19 +04:00
* object . This is the minimum of the full length requested ( len ) or
2009-10-30 22:57:30 +03:00
* the remainder of the current stripe being written to .
*/
2012-11-14 19:38:19 +04:00
* oxlen = min_t ( u64 , len , su - su_offset ) ;
2009-10-06 22:31:10 +04:00
dout ( " obj extent %llu~%llu \n " , * oxoff , * oxlen ) ;
2012-09-25 07:59:48 +04:00
return 0 ;
invalid :
dout ( " invalid layout \n " ) ;
* ono = 0 ;
* oxoff = 0 ;
* oxlen = 0 ;
return - EINVAL ;
2009-10-06 22:31:10 +04:00
}
2010-04-07 02:14:15 +04:00
EXPORT_SYMBOL ( ceph_calc_file_object_mapping ) ;
2009-10-06 22:31:10 +04:00
/*
2014-01-27 19:40:19 +04:00
* Calculate mapping of a ( oloc , oid ) pair to a PG . Should only be
* called with target ' s ( oloc , oid ) , since tiering isn ' t taken into
* account .
2009-10-06 22:31:10 +04:00
*/
2014-01-27 19:40:19 +04:00
int ceph_oloc_oid_to_pg ( struct ceph_osdmap * osdmap ,
struct ceph_object_locator * oloc ,
struct ceph_object_id * oid ,
struct ceph_pg * pg_out )
2009-10-06 22:31:10 +04:00
{
2014-01-27 19:40:19 +04:00
struct ceph_pg_pool_info * pi ;
2009-10-06 22:31:10 +04:00
2014-01-27 19:40:19 +04:00
pi = __lookup_pg_pool ( & osdmap - > pg_pools , oloc - > pool ) ;
if ( ! pi )
2010-02-17 02:55:03 +03:00
return - EIO ;
2009-10-06 22:31:10 +04:00
2014-01-27 19:40:19 +04:00
pg_out - > pool = oloc - > pool ;
pg_out - > seed = ceph_str_hash ( pi - > object_hash , oid - > name ,
oid - > name_len ) ;
dout ( " %s '%.*s' pgid %llu.%x \n " , __func__ , oid - > name_len , oid - > name ,
pg_out - > pool , pg_out - > seed ) ;
2009-10-06 22:31:10 +04:00
return 0 ;
}
2014-01-27 19:40:19 +04:00
EXPORT_SYMBOL ( ceph_oloc_oid_to_pg ) ;
2009-10-06 22:31:10 +04:00
2014-01-31 19:54:26 +04:00
static int do_crush ( struct ceph_osdmap * map , int ruleno , int x ,
int * result , int result_max ,
const __u32 * weight , int weight_max )
2013-12-24 23:19:24 +04:00
{
2014-01-31 19:54:26 +04:00
int r ;
BUG_ON ( result_max > CEPH_PG_MAX_SIZE ) ;
mutex_lock ( & map - > crush_scratch_mutex ) ;
r = crush_do_rule ( map - > crush , ruleno , x , result , result_max ,
weight , weight_max , map - > crush_scratch_ary ) ;
mutex_unlock ( & map - > crush_scratch_mutex ) ;
2013-12-24 23:19:24 +04:00
2014-01-31 19:54:26 +04:00
return r ;
2013-12-24 23:19:24 +04:00
}
2009-10-06 22:31:10 +04:00
/*
* Calculate raw osd vector for the given pgid . Return pointer to osd
* array , or NULL on failure .
*/
2013-02-23 22:38:16 +04:00
static int * calc_pg_raw ( struct ceph_osdmap * osdmap , struct ceph_pg pgid ,
2009-10-06 22:31:10 +04:00
int * osds , int * num )
{
struct ceph_pg_mapping * pg ;
struct ceph_pg_pool_info * pool ;
int ruleno ;
2013-02-26 22:39:09 +04:00
int r ;
u32 pps ;
2009-10-06 22:31:10 +04:00
2013-02-26 22:39:09 +04:00
pool = __lookup_pg_pool ( & osdmap - > pg_pools , pgid . pool ) ;
2011-09-28 21:08:27 +04:00
if ( ! pool )
return NULL ;
2009-10-06 22:31:10 +04:00
/* pg_temp? */
2013-02-26 22:39:09 +04:00
pgid . seed = ceph_stable_mod ( pgid . seed , pool - > pg_num ,
2013-08-29 04:17:29 +04:00
pool - > pg_num_mask ) ;
2010-02-17 02:53:32 +03:00
pg = __lookup_pg_mapping ( & osdmap - > pg_temp , pgid ) ;
if ( pg ) {
* num = pg - > len ;
return pg - > osds ;
2009-10-06 22:31:10 +04:00
}
/* crush */
2013-02-23 22:41:09 +04:00
ruleno = crush_find_rule ( osdmap - > crush , pool - > crush_ruleset ,
pool - > type , pool - > size ) ;
2009-10-06 22:31:10 +04:00
if ( ruleno < 0 ) {
2013-02-26 22:39:09 +04:00
pr_err ( " no crush rule pool %lld ruleset %d type %d size %d \n " ,
pgid . pool , pool - > crush_ruleset , pool - > type ,
2013-02-23 22:41:09 +04:00
pool - > size ) ;
2009-10-06 22:31:10 +04:00
return NULL ;
}
2013-02-26 22:39:09 +04:00
if ( pool - > flags & CEPH_POOL_FLAG_HASHPSPOOL ) {
/* hash pool id and seed sothat pool PGs do not overlap */
pps = crush_hash32_2 ( CRUSH_HASH_RJENKINS1 ,
ceph_stable_mod ( pgid . seed , pool - > pgp_num ,
pool - > pgp_num_mask ) ,
pgid . pool ) ;
} else {
/*
* legacy ehavior : add ps and pool together . this is
* not a great approach because the PGs from each pool
* will overlap on top of each other : 0.5 = = 1.4 = =
* 2.3 = = . . .
*/
pps = ceph_stable_mod ( pgid . seed , pool - > pgp_num ,
pool - > pgp_num_mask ) +
( unsigned ) pgid . pool ;
}
2014-01-31 19:54:26 +04:00
r = do_crush ( osdmap , ruleno , pps , osds , min_t ( int , pool - > size , * num ) ,
osdmap - > osd_weight , osdmap - > max_osd ) ;
2012-05-08 02:37:23 +04:00
if ( r < 0 ) {
2013-02-26 22:39:09 +04:00
pr_err ( " error %d from crush rule: pool %lld ruleset %d type %d "
" size %d \n " , r , pgid . pool , pool - > crush_ruleset ,
2013-02-23 22:41:09 +04:00
pool - > type , pool - > size ) ;
2012-05-08 02:37:23 +04:00
return NULL ;
}
* num = r ;
2009-10-06 22:31:10 +04:00
return osds ;
}
2010-05-10 21:24:48 +04:00
/*
* Return acting set for given pgid .
*/
2013-02-23 22:38:16 +04:00
int ceph_calc_pg_acting ( struct ceph_osdmap * osdmap , struct ceph_pg pgid ,
2010-05-10 21:24:48 +04:00
int * acting )
{
int rawosds [ CEPH_PG_MAX_SIZE ] , * osds ;
int i , o , num = CEPH_PG_MAX_SIZE ;
osds = calc_pg_raw ( osdmap , pgid , rawosds , & num ) ;
if ( ! osds )
return - 1 ;
/* primary is first up osd */
o = 0 ;
for ( i = 0 ; i < num ; i + + )
if ( ceph_osd_is_up ( osdmap , osds [ i ] ) )
acting [ o + + ] = osds [ i ] ;
return o ;
}
2009-10-06 22:31:10 +04:00
/*
* Return primary osd for given pgid , or - 1 if none .
*/
2013-02-23 22:38:16 +04:00
int ceph_calc_pg_primary ( struct ceph_osdmap * osdmap , struct ceph_pg pgid )
2009-10-06 22:31:10 +04:00
{
2010-05-10 21:24:48 +04:00
int rawosds [ CEPH_PG_MAX_SIZE ] , * osds ;
int i , num = CEPH_PG_MAX_SIZE ;
2009-10-06 22:31:10 +04:00
osds = calc_pg_raw ( osdmap , pgid , rawosds , & num ) ;
if ( ! osds )
return - 1 ;
/* primary is first up osd */
for ( i = 0 ; i < num ; i + + )
2010-05-10 21:24:48 +04:00
if ( ceph_osd_is_up ( osdmap , osds [ i ] ) )
2009-10-06 22:31:10 +04:00
return osds [ i ] ;
return - 1 ;
}
2010-04-07 02:14:15 +04:00
EXPORT_SYMBOL ( ceph_calc_pg_primary ) ;