Skip to content

Conversation

@beauregardc
Copy link

fix a memory leak caused by not freeing existing RTMD when duplicating RTMD from something else. It doesn't happen for every product, but the 307091 template appears to trigger it reliably. The memcheck call stack for the leak tends to look something like:

==543602== 80 bytes in 20 blocks are indirectly lost in loss record 1 of 2
==543602== at 0x483DD99: calloc (in /usr/lib/x86_64-linux-gnu/valgrind/vgpreload_memcheck-amd64-linux.so)
==543602== by 0x4856E74: bufr_create_rtmd (bufr_meta.c:63)
==543602== by 0x4856EDA: bufr_duplicate_rtmd (bufr_meta.c:100)
==543602== by 0x485A3A7: bufr_copy_descriptor (bufr_desc.c:282)
==543602== by 0x485A1C3: bufr_dupl_descriptor (bufr_desc.c:206)
==543602== by 0x485CF5F: bufr_repl_descriptors (bufr_sequence.c:599)
==543602== by 0x485C3B2: bufr_expand_node_descriptor (bufr_sequence.c:354)
==543602== by 0x485BD56: bufr_expand_list (bufr_sequence.c:201)
==543602== by 0x485C167: bufr_expand_desc (bufr_sequence.c:311)
==543602== by 0x485C8F4: bufr_expand_node_descriptor (bufr_sequence.c:438)
==543602== by 0x485BD56: bufr_expand_list (bufr_sequence.c:201)
==543602== by 0x485D27B: bufr_repl_descriptors (bufr_sequence.c:671)
==543602== by 0x485C6C8: bufr_expand_node_descriptor (bufr_sequence.c:402)
==543602== by 0x486F81A: bufr_decode_message (bufr_dataset.c:2413)
==543602== by 0x10B9AD: run_decoder (bufr_decoder.c:365)
==543602== by 0x10B4F5: main (bufr_decoder.c:245)
==543602==
==543602== 1,200 (1,120 direct, 80 indirect) bytes in 20 blocks are definitely lost in loss record 2 of 2
==543602== at 0x483DD99: calloc (in /usr/lib/x86_64-linux-gnu/valgrind/vgpreload_memcheck-amd64-linux.so)
==543602== by 0x4856E40: bufr_create_rtmd (bufr_meta.c:58)
==543602== by 0x4856EDA: bufr_duplicate_rtmd (bufr_meta.c:100)
==543602== by 0x485A3A7: bufr_copy_descriptor (bufr_desc.c:282)
==543602== by 0x485A1C3: bufr_dupl_descriptor (bufr_desc.c:206)
==543602== by 0x485CF5F: bufr_repl_descriptors (bufr_sequence.c:599)
==543602== by 0x485C3B2: bufr_expand_node_descriptor (bufr_sequence.c:354)
==543602== by 0x485BD56: bufr_expand_list (bufr_sequence.c:201)
==543602== by 0x485C167: bufr_expand_desc (bufr_sequence.c:311)
==543602== by 0x485C8F4: bufr_expand_node_descriptor (bufr_sequence.c:438)
==543602== by 0x485BD56: bufr_expand_list (bufr_sequence.c:201)
==543602== by 0x485D27B: bufr_repl_descriptors (bufr_sequence.c:671)
==543602== by 0x485C6C8: bufr_expand_node_descriptor (bufr_sequence.c:402)
==543602== by 0x486F81A: bufr_decode_message (bufr_dataset.c:2413)
==543602== by 0x10B9AD: run_decoder (bufr_decoder.c:365)
==543602== by 0x10B4F5: main (bufr_decoder.c:245)

RTMD from something else. It doesn't happen for every product, but
the 307091 template appears to trigger it reliably. The memcheck call stack for
the leak tends to look something like:

==543602== 80 bytes in 20 blocks are indirectly lost in loss record 1 of 2
==543602==    at 0x483DD99: calloc (in /usr/lib/x86_64-linux-gnu/valgrind/vgpreload_memcheck-amd64-linux.so)
==543602==    by 0x4856E74: bufr_create_rtmd (bufr_meta.c:63)
==543602==    by 0x4856EDA: bufr_duplicate_rtmd (bufr_meta.c:100)
==543602==    by 0x485A3A7: bufr_copy_descriptor (bufr_desc.c:282)
==543602==    by 0x485A1C3: bufr_dupl_descriptor (bufr_desc.c:206)
==543602==    by 0x485CF5F: bufr_repl_descriptors (bufr_sequence.c:599)
==543602==    by 0x485C3B2: bufr_expand_node_descriptor (bufr_sequence.c:354)
==543602==    by 0x485BD56: bufr_expand_list (bufr_sequence.c:201)
==543602==    by 0x485C167: bufr_expand_desc (bufr_sequence.c:311)
==543602==    by 0x485C8F4: bufr_expand_node_descriptor (bufr_sequence.c:438)
==543602==    by 0x485BD56: bufr_expand_list (bufr_sequence.c:201)
==543602==    by 0x485D27B: bufr_repl_descriptors (bufr_sequence.c:671)
==543602==    by 0x485C6C8: bufr_expand_node_descriptor (bufr_sequence.c:402)
==543602==    by 0x486F81A: bufr_decode_message (bufr_dataset.c:2413)
==543602==    by 0x10B9AD: run_decoder (bufr_decoder.c:365)
==543602==    by 0x10B4F5: main (bufr_decoder.c:245)
==543602==
==543602== 1,200 (1,120 direct, 80 indirect) bytes in 20 blocks are definitely lost in loss record 2 of 2
==543602==    at 0x483DD99: calloc (in /usr/lib/x86_64-linux-gnu/valgrind/vgpreload_memcheck-amd64-linux.so)
==543602==    by 0x4856E40: bufr_create_rtmd (bufr_meta.c:58)
==543602==    by 0x4856EDA: bufr_duplicate_rtmd (bufr_meta.c:100)
==543602==    by 0x485A3A7: bufr_copy_descriptor (bufr_desc.c:282)
==543602==    by 0x485A1C3: bufr_dupl_descriptor (bufr_desc.c:206)
==543602==    by 0x485CF5F: bufr_repl_descriptors (bufr_sequence.c:599)
==543602==    by 0x485C3B2: bufr_expand_node_descriptor (bufr_sequence.c:354)
==543602==    by 0x485BD56: bufr_expand_list (bufr_sequence.c:201)
==543602==    by 0x485C167: bufr_expand_desc (bufr_sequence.c:311)
==543602==    by 0x485C8F4: bufr_expand_node_descriptor (bufr_sequence.c:438)
==543602==    by 0x485BD56: bufr_expand_list (bufr_sequence.c:201)
==543602==    by 0x485D27B: bufr_repl_descriptors (bufr_sequence.c:671)
==543602==    by 0x485C6C8: bufr_expand_node_descriptor (bufr_sequence.c:402)
==543602==    by 0x486F81A: bufr_decode_message (bufr_dataset.c:2413)
==543602==    by 0x10B9AD: run_decoder (bufr_decoder.c:365)
==543602==    by 0x10B4F5: main (bufr_decoder.c:245)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant