raw2trace assert

[Prasun Ratn]

I am seeing this assert:

[drmemtrace]: Mapped module 8 @0x7f29ce4c5000-0x7f29ce65b000 (-0x7f29ce510000 segment) = /home/prasun/dynamorio/build/lib64/release/libdynamorio.so
[drmemtrace]: Secondary segment: module 9 seg 0x7f29ce510000-0x7f29ce5fe000 = /home/prasun/dynamorio/build/lib64/release/libdynamorio.so
[drmemtrace]: Secondary segment: module 10 seg 0x7f29ce5fe000-0x7f29ce648000 = /home/prasun/dynamorio/build/lib64/release/libdynamorio.so
[drmemtrace]: Secondary segment: module 11 seg 0x7f29ce649000-0x7f29ce6e2000 = /home/prasun/dynamorio/build/lib64/release/libdynamorio.so
ASSERT FAILURE: /home/prasun/dynamorio/dynamorio/clients/drcachesim/tracer/raw2trace.cpp:371: off_end || info.start - modvec_[info.containing_index].orig_seg_base + info.size <= modvec_[info.containing_index].total_map_size ()


 id, containing_id, start, end, entry, offset, preferred_base, (custom fields), path
  8,   8, 0x000014d184a99000, 0x000014d184ae4000, 0x000014d184ba7875, 0000000000000000, 0x0000000071000000, v#1,0, /home/prasun/dynamorio/build/lib64/release/libdynamorio.so
  9,   8, 0x000014d184ae4000, 0x000014d184bd2000, 0x000014d184ba7875, 000000000004b000, 0x000000007104b000, v#1,0, /home/prasun/dynamorio/build/lib64/release/libdynamorio.so
 10,   8, 0x000014d184bd2000, 0x000014d184c1c000, 0x000014d184ba7875, 0000000000139000, 0x0000000071139000, v#1,0, /home/prasun/dynamorio/build/lib64/release/libdynamorio.so
 11,   8, 0x000014d184c1d000, 0x000014d184cb6000, 0x000014d184ba7875, 0000000000183ae0, 0x0000000071184000, v#1,0, /home/prasun/dynamorio/build/lib64/release/libdynamorio.so

It looks like this is occurring because the size of module 8 does not match across raw2trace run and the modules.log entry.

>>> print 0x000014d184ae4000-0x000014d184a99000
307200
>>> print 0x7f29ce65b000-0x7f29ce4c5000
1662976

Any idea why this could happen? I don't think the libdynamorio.so file has changed since tracing (but cannot be 100% sure).

Thanks

Prasun

[Nahome Bete]

Hi Prasun,

Can you provide us more details following the template at https://github.com/DynamoRIO/dynamorio/issues/new?assignees=&labels=&template=bug_report.md&title=%3CEDIT+ME%3E+%28following+https%3A%2F%2F%2Fdynamorio.org%2Fissues+with+one+of+the+CRASH%2C+APP+CRASH%2C+HANG%2C+or+ASSERT+keywords%29 so that we can reproduce it locally?

[Derek Bruening]

The only place I've seen that assert is from segment alignment issues: https://github.com/DynamoRIO/dynamorio/issues/4731, https://github.com/DynamoRIO/dynamorio/issues/4737.

--
You received this message because you are subscribed to the Google Groups "DynamoRIO Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dynamorio-use...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dynamorio-users/3e292bab-ec25-4f9b-a99f-e8a28318b6fbn%40googlegroups.com.

[Prasun Ratn]

Derek, that doesn't seem to be the case here, correct?

Nahome, I don't think I have enough to file a bug yet. Just trying to understand what the assert is saying.

[Derek Bruening]

If it reproduces you can verify whether it is the same file?

[Prasun Ratn]

Yeah it did not reproduce when I tried. Is there a way to enable build id logging in modules.log?

[Derek Bruening]

We add them to all ours using the additional field interface.  Seems reasonable to add upstream to Linux since most distros these days have them in system libraries; not all libraries have them, though, like checksums and timestamps and other fields, so they are not a panacea.

To view this discussion on the web visit https://groups.google.com/d/msgid/dynamorio-users/52e85d41-6678-4de5-ac7b-4cbf7f5be5e0n%40googlegroups.com.

[Derek Bruening]

https://github.com/DynamoRIO/dynamorio/issues/5550

[Prasun Ratn]

Thanks Derek.