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

top: finally circumvent that minor libnuma memory leak

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.

[ this commit parallels a patch in our newlib branch ]

Signed-off-by: Jim Warner <james.warner@comcast.net>
This commit is contained in:
Jim Warner 2016-09-12 16:16:16 -05:00 committed by Craig Small
parent 24fd2605c5
commit 18b1a887b7
2 changed files with 14 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
NEWS
View File

@ -4,6 +4,7 @@ procps-ng-NEXT
* library: dont use vm_min_free on non Linux Debian #831396
* library: dont use SIGPWR on FreeBSD Debian #832148
* pmap: fix duplicate output line under '-x' option Redhat #1374061
* top: eliminated minor libnuma memory leak
procps-ng-3.3.12
----------------

15
top/top.c
View File

@ -552,7 +552,16 @@ static void bye_bye (const char *str) {
#endif // end: OFF_HST_HASH
#ifndef NUMA_DISABLE
if (Libnuma_handle) dlclose(Libnuma_handle);
/* note: we'll skip a dlcose() to avoid the following libnuma memory
* 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 (Libnuma_handle)
// dlclose(Libnuma_handle);
#endif
if (str) {
fputs(str, stderr);
@ -1717,7 +1726,7 @@ static FLD_t Fieldstab[] = {
EU_UED, L_NONE - natural outgrowth of 'stat()' in readproc (euid)
EU_CPU, L_stat - never filled by libproc, but requires times (pcpu)
EU_CMD, L_stat - may yet require L_CMDLINE in calibrate_fields (cmd/cmdline)
L_EITHER - must L_status, else L_stat == 64-bit math (__udivdi3) on 32-bit !
L_EITHER - favor L_stat (L_status == ++cost of gpref & hash scheme)
.width .scale .align .sort .lflg
------ ------ -------- -------- -------- */
@ -3318,6 +3327,8 @@ static void before (char *me) {
if (Numa_max_node && Numa_node_of_cpu)
Numa_node_tot = Numa_max_node() + 1;
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(Libnuma_handle);
Libnuma_handle = NULL;
}