emo/procps
1
0
Fork 0
Code Issues Pull Requests Packages Releases Activity

library: finally circumvent libnuma memory leak <stat>

Browse Source 
Still unhappy with a minor memory leak associated with
libnuma, I experimented with omitting the dlclose that
was issued at module's end. For some reason which will
remain a mystery, the valgrind leak then went bye-bye.

So this patch just omits one use of dlclose and relies
on whatever kernel magic is at work to free the memory
when each process ends. We kept, however, the original
code (now commented-out) to serve as a future caution.

There remains one potential (but unlikely) dlclose use
near the original dlopen. But there will be no leak as
that 'numa_node_of_cpu' will not yet have been called.
This seems to be the culprit that triggers such leaks.

None of this libnuma shit would likely have come close
to hitting our fan had the numa developers provided us
with 'new' and 'unref' functions like our newlib does.

Signed-off-by: Jim Warner <james.warner@comcast.net>
This commit is contained in:
Jim Warner 2016-09-12 15:15:15 -05:00 committed by Craig Small
parent a8f7e44233
commit 91d47123f2
1 changed files with 13 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
proc/stat.c
View File

@ -489,15 +489,7 @@ static int stat_make_numa_hist (
nod_ptr->new.irq += cpu_ptr->new.irq; nod_ptr->old.irq += cpu_ptr->old.irq; nod_ptr->new.irq += cpu_ptr->new.irq; nod_ptr->old.irq += cpu_ptr->old.irq;
nod_ptr->new.sirq += cpu_ptr->new.sirq; nod_ptr->old.sirq += cpu_ptr->old.sirq; nod_ptr->new.sirq += cpu_ptr->new.sirq; nod_ptr->old.sirq += cpu_ptr->old.sirq;
nod_ptr->new.stolen += cpu_ptr->new.stolen; nod_ptr->old.stolen += cpu_ptr->old.stolen; nod_ptr->new.stolen += cpu_ptr->new.stolen; nod_ptr->old.stolen += cpu_ptr->old.stolen;
/*
* note: the above call to 'our_node_of_cpu' will produce a modest
* memory leak summarized as:
* ==1234== LEAK SUMMARY:
* ==1234== definitely lost: 512 bytes in 1 blocks
* ==1234== indirectly lost: 48 bytes in 2 blocks
* ==1234== ...
* [ thanks very much libnuma, for all the pain you've caused us ]
*/
cpu_ptr->numa_node = node; cpu_ptr->numa_node = node;
nod_ptr->id = node; nod_ptr->id = node;
} }
@ -867,6 +859,8 @@ PROCPS_EXPORT int procps_stat_new (
if (p->our_max_node && p->our_node_of_cpu) if (p->our_max_node && p->our_node_of_cpu)
p->nodes.total = p->our_max_node() + 1; p->nodes.total = p->our_max_node() + 1;
else { else {
// this dlclose is safe - we've yet to call numa_node_of_cpu
// ( there's one other dlclose which has now been disabled )
dlclose(p->libnuma_handle); dlclose(p->libnuma_handle);
p->libnuma_handle = NULL; p->libnuma_handle = NULL;
} }
@ -934,8 +928,16 @@ PROCPS_EXPORT int procps_stat_unref (
#ifndef NUMA_DISABLE #ifndef NUMA_DISABLE
#ifndef PRETEND_NUMA #ifndef PRETEND_NUMA
if ((*info)->libnuma_handle) /* note: we'll skip a dlcose() to avoid the following libnuma memory
dlclose((*info)->libnuma_handle); * leak which is triggered after a call to numa_node_of_cpu():
* ==1234== LEAK SUMMARY:
* ==1234== definitely lost: 512 bytes in 1 blocks
* ==1234== indirectly lost: 48 bytes in 2 blocks
* ==1234== ...
* [ thanks very much libnuma, for all the pain you've caused ]
*/
// if ((*info)->libnuma_handle)
// dlclose((*info)->libnuma_handle);
#endif #endif
#endif #endif
free(*info); free(*info);