Valgrind Uninitialized Error about libmseed HOT 5 CLOSED

No worries. Thanks for the report. Valgrind is awesome for this kind of thing, keeps C honest until we wholesale move to Rust...

from libmseed.

Thanks. Is this on current master branch or some other tagged release?

from libmseed.

Hey Chad, this is on the current master. I think the issue may have existed for at least a few months on libmseed v3. Unfortunately, I was tracking a problem with my codes, valgrind complained, so I just pulled the latest from the master hoping the problem would magically disappear. I didn't note my previous libmseed3 version. Sorry!

from libmseed.

Related. Apologies for a terrible bug report, this is from memory.
When:
ms3_readtracelist( without UNPACKDATA and then again with UNPACKDATA
I think I saw datasamples alloc'd twice, leaving a dangling pointer from the first call.... or maybe some other memory bug that valgrind should easily catch.
I meant to make a proper case report, and PR, but alas, time.

from libmseed.

I'm pretty certain that the lack of initialisation of fpos on line 675 of fileutils.c has been causing errors for me, and that initialising it to 0 fixes those issues.

For background, I'm wrapping libmseed in a Julia package, and I have several tests calling ms3_readtracelist a total of I guess ~100 times. My tests were failing on certain machine and architecture combinations, with files not being read properly. (See e.g. https://github.com/anowacki/LibMseed.jl/runs/5971465915?check_suite_focus=true for error messages logged by libmseed amongst all the Julia stuff.) In short, calls to ms3_readtracelist for data which had just been written by msr3_writemseed and existing test data were returning a MS_NOTSEED error, even though the files were valid miniSEED on disk, and other OS-architecture combinations were working.

Initialising fpos to 0 as suggested by @bakerb845 fixes these problems for me (e.g., https://github.com/anowacki/LibMseed.jl/runs/5981051328?check_suite_focus=true), so I wonder if you might want to include this change in a future version of libmseed soon? I have submitted a PR to this effect (#80), but please do feel free to close that and deal with this as you see fit.

from libmseed.